talksik/mediapipe
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

Project import generated by Copybara.

Browse Source 
GitOrigin-RevId: 373e3ac1e5839befd95bf7d73ceff3c5f1171969
This commit is contained in:
MediaPipe Team
2021-10-06 14:27:49 -07:00
committed by jqtang
parent 137e1cc763
commit 33d683c671
153 changed files with 7871 additions and 1349 deletions
Download Patch File Download Diff File Expand all files Collapse all files
mediapipe/modules/face_detection/BUILD
+24
View File
@@ -117,6 +117,30 @@ mediapipe_simple_subgraph(
],
)
mediapipe_simple_subgraph(
name = "face_detection_short_range_image",
graph = "face_detection_short_range_image.pbtxt",
register_as = "FaceDetectionShortRangeImage",
deps = [
":face_detection_short_range_common",
"//mediapipe/calculators/core:flow_limiter_calculator",
"//mediapipe/calculators/tensor:image_to_tensor_calculator",
"//mediapipe/calculators/tensor:inference_calculator",
],
)
mediapipe_simple_subgraph(
name = "face_detection_full_range_image",
graph = "face_detection_full_range_image.pbtxt",
register_as = "FaceDetectionFullRangeImage",
deps = [
":face_detection_full_range_common",
"//mediapipe/calculators/core:flow_limiter_calculator",
"//mediapipe/calculators/tensor:image_to_tensor_calculator",
"//mediapipe/calculators/tensor:inference_calculator",
],
)
exports_files(
srcs = [
"face_detection_full_range.tflite",
mediapipe/modules/face_detection/face_detection_full_range_image.pbtxt
+86
View File
@@ -0,0 +1,86 @@
# MediaPipe graph to detect faces. (GPU/CPU input, and inference is executed on
# GPU.)
#
# It is required that "face_detection_full_range_sparse.tflite" is available at
# "mediapipe/modules/face_detection/face_detection_full_range_sparse.tflite"
# path during execution.
type: "FaceDetectionFullRangeImage"
# Image. (Image)
input_stream: "IMAGE:image"
# The throttled input image. (Image)
output_stream: "IMAGE:throttled_image"
# Detected faces. (std::vector<Detection>)
# NOTE: there will not be an output packet in the DETECTIONS stream for this
# particular timestamp if none of faces detected. However, the MediaPipe
# framework will internally inform the downstream calculators of the absence of
# this packet so that they don't wait for it unnecessarily.
output_stream: "DETECTIONS:detections"
node {
calculator: "FlowLimiterCalculator"
input_stream: "image"
input_stream: "FINISHED:detections"
input_stream_info: {
tag_index: "FINISHED"
back_edge: true
}
output_stream: "throttled_image"
options: {
[mediapipe.FlowLimiterCalculatorOptions.ext] {
max_in_flight: 1
max_in_queue: 1
}
}
}
# Transforms the input image into a 128x128 tensor while keeping the aspect
# ratio (what is expected by the corresponding face detection model), resulting
# in potential letterboxing in the transformed image.
node: {
calculator: "ImageToTensorCalculator"
input_stream: "IMAGE:throttled_image"
output_stream: "TENSORS:input_tensors"
output_stream: "MATRIX:transform_matrix"
options: {
[mediapipe.ImageToTensorCalculatorOptions.ext] {
output_tensor_width: 192
output_tensor_height: 192
keep_aspect_ratio: true
output_tensor_float_range {
min: -1.0
max: 1.0
}
border_mode: BORDER_ZERO
gpu_origin: CONVENTIONAL
}
}
}
# Runs a TensorFlow Lite model on GPU that takes an image tensor and outputs a
# vector of tensors representing, for instance, detection boxes/keypoints and
# scores.
# TODO: Use GraphOptions to modify the delegate field to be
# `delegate { xnnpack {} }` for the CPU only use cases.
node {
calculator: "InferenceCalculator"
input_stream: "TENSORS:input_tensors"
output_stream: "TENSORS:detection_tensors"
options: {
[mediapipe.InferenceCalculatorOptions.ext] {
model_path: "mediapipe/modules/face_detection/face_detection_full_range_sparse.tflite"
#
delegate: { gpu { use_advanced_gpu_api: true } }
}
}
}
# Performs tensor post processing to generate face detections.
node {
calculator: "FaceDetectionFullRangeCommon"
input_stream: "TENSORS:detection_tensors"
input_stream: "MATRIX:transform_matrix"
output_stream: "DETECTIONS:detections"
}
mediapipe/modules/face_detection/face_detection_short_range_image.pbtxt
+94
View File
@@ -0,0 +1,94 @@
# MediaPipe graph to detect faces. (GPU/CPU input, and inference is executed on
# GPU.)
#
# It is required that "face_detection_short_range.tflite" is available at
# "mediapipe/modules/face_detection/face_detection_short_range.tflite"
# path during execution.
#
# EXAMPLE:
# node {
# calculator: "FaceDetectionShortRangeCpu"
# input_stream: "IMAGE:image"
# output_stream: "DETECTIONS:face_detections"
# }
type: "FaceDetectionShortRangeCpu"
# Image. (Image)
input_stream: "IMAGE:image"
# The throttled input image. (Image)
output_stream: "IMAGE:throttled_image"
# Detected faces. (std::vector<Detection>)
# NOTE: there will not be an output packet in the DETECTIONS stream for this
# particular timestamp if none of faces detected. However, the MediaPipe
# framework will internally inform the downstream calculators of the absence of
# this packet so that they don't wait for it unnecessarily.
output_stream: "DETECTIONS:detections"
node {
calculator: "FlowLimiterCalculator"
input_stream: "image"
input_stream: "FINISHED:detections"
input_stream_info: {
tag_index: "FINISHED"
back_edge: true
}
output_stream: "throttled_image"
options: {
[mediapipe.FlowLimiterCalculatorOptions.ext] {
max_in_flight: 1
max_in_queue: 1
}
}
}
# Transforms the input image into a 128x128 tensor while keeping the aspect
# ratio (what is expected by the corresponding face detection model), resulting
# in potential letterboxing in the transformed image.
node: {
calculator: "ImageToTensorCalculator"
input_stream: "IMAGE:throttled_image"
output_stream: "TENSORS:input_tensors"
output_stream: "MATRIX:transform_matrix"
options: {
[mediapipe.ImageToTensorCalculatorOptions.ext] {
output_tensor_width: 128
output_tensor_height: 128
keep_aspect_ratio: true
output_tensor_float_range {
min: -1.0
max: 1.0
}
border_mode: BORDER_ZERO
gpu_origin: CONVENTIONAL
}
}
}
# Runs a TensorFlow Lite model on GPU that takes an image tensor and outputs a
# vector of tensors representing, for instance, detection boxes/keypoints and
# scores.
# TODO: Use GraphOptions to modify the delegate field to be
# `delegate { xnnpack {} }` for the CPU only use cases.
node {
calculator: "InferenceCalculator"
input_stream: "TENSORS:input_tensors"
output_stream: "TENSORS:detection_tensors"
options: {
[mediapipe.InferenceCalculatorOptions.ext] {
model_path: "mediapipe/modules/face_detection/face_detection_short_range.tflite"
#
delegate: { gpu { use_advanced_gpu_api: true } }
}
}
}
# Performs tensor post processing to generate face detections.
node {
calculator: "FaceDetectionShortRangeCommon"
input_stream: "TENSORS:detection_tensors"
input_stream: "MATRIX:transform_matrix"
output_stream: "DETECTIONS:detections"
}
mediapipe/modules/face_landmark/BUILD
+45
View File
@@ -26,14 +26,19 @@ mediapipe_simple_subgraph(
graph = "face_landmark_cpu.pbtxt",
register_as = "FaceLandmarkCpu",
deps = [
":face_landmarks_model_loader",
":tensors_to_face_landmarks",
":tensors_to_face_landmarks_with_attention",
"//mediapipe/calculators/core:gate_calculator",
"//mediapipe/calculators/core:split_vector_calculator",
"//mediapipe/calculators/tensor:image_to_tensor_calculator",
"//mediapipe/calculators/tensor:inference_calculator",
"//mediapipe/calculators/tensor:tensors_to_floats_calculator",
"//mediapipe/calculators/tensor:tensors_to_landmarks_calculator",
"//mediapipe/calculators/tflite:tflite_custom_op_resolver_calculator",
"//mediapipe/calculators/util:landmark_projection_calculator",
"//mediapipe/calculators/util:thresholding_calculator",
"//mediapipe/framework/tool:switch_container",
],
)
@@ -42,14 +47,19 @@ mediapipe_simple_subgraph(
graph = "face_landmark_gpu.pbtxt",
register_as = "FaceLandmarkGpu",
deps = [
":face_landmarks_model_loader",
":tensors_to_face_landmarks",
":tensors_to_face_landmarks_with_attention",
"//mediapipe/calculators/core:gate_calculator",
"//mediapipe/calculators/core:split_vector_calculator",
"//mediapipe/calculators/tensor:image_to_tensor_calculator",
"//mediapipe/calculators/tensor:inference_calculator",
"//mediapipe/calculators/tensor:tensors_to_floats_calculator",
"//mediapipe/calculators/tensor:tensors_to_landmarks_calculator",
"//mediapipe/calculators/tflite:tflite_custom_op_resolver_calculator",
"//mediapipe/calculators/util:landmark_projection_calculator",
"//mediapipe/calculators/util:thresholding_calculator",
"//mediapipe/framework/tool:switch_container",
],
)
@@ -101,6 +111,7 @@ mediapipe_simple_subgraph(
register_as = "FaceLandmarkFrontCpuImage",
deps = [
":face_landmark_front_cpu",
"//mediapipe/calculators/core:flow_limiter_calculator",
"//mediapipe/calculators/image:image_transformation_calculator",
"//mediapipe/calculators/util:from_image_calculator",
],
@@ -112,6 +123,7 @@ mediapipe_simple_subgraph(
register_as = "FaceLandmarkFrontGpuImage",
deps = [
":face_landmark_front_gpu",
"//mediapipe/calculators/core:flow_limiter_calculator",
"//mediapipe/calculators/image:image_transformation_calculator",
"//mediapipe/calculators/util:from_image_calculator",
],
@@ -120,6 +132,7 @@ mediapipe_simple_subgraph(
exports_files(
srcs = [
"face_landmark.tflite",
"face_landmark_with_attention.tflite",
],
)
@@ -143,3 +156,35 @@ mediapipe_simple_subgraph(
"//mediapipe/calculators/util:rect_transformation_calculator",
],
)
mediapipe_simple_subgraph(
name = "face_landmarks_model_loader",
graph = "face_landmarks_model_loader.pbtxt",
register_as = "FaceLandmarksModelLoader",
deps = [
"//mediapipe/calculators/core:constant_side_packet_calculator",
"//mediapipe/calculators/tflite:tflite_model_calculator",
"//mediapipe/calculators/util:local_file_contents_calculator",
"//mediapipe/framework/tool:switch_container",
],
)
mediapipe_simple_subgraph(
name = "tensors_to_face_landmarks",
graph = "tensors_to_face_landmarks.pbtxt",
register_as = "TensorsToFaceLandmarks",
deps = [
"//mediapipe/calculators/tensor:tensors_to_landmarks_calculator",
],
)
mediapipe_simple_subgraph(
name = "tensors_to_face_landmarks_with_attention",
graph = "tensors_to_face_landmarks_with_attention.pbtxt",
register_as = "TensorsToFaceLandmarksWithAttention",
deps = [
"//mediapipe/calculators/core:split_vector_calculator",
"//mediapipe/calculators/tensor:tensors_to_landmarks_calculator",
"//mediapipe/calculators/util:landmarks_refinement_calculator",
],
)
mediapipe/modules/face_landmark/face_landmark.tflite
BIN
View File
Binary file not shown.
mediapipe/modules/face_landmark/face_landmark_cpu.pbtxt
+64 -14
View File
@@ -3,13 +3,18 @@
#
# It is required that "face_landmark.tflite" is available at
# "mediapipe/modules/face_landmark/face_landmark.tflite"
# path during execution.
# path during execution if `with_attention` is not set or set to `false`.
#
# It is required that "face_landmark_with_attention.tflite" is available at
# "mediapipe/modules/face_landmark/face_landmark_with_attention.tflite"
# path during execution if `with_attention` is set to `true`.
#
# EXAMPLE:
# node {
# calculator: "FaceLandmarkCpu"
# input_stream: "IMAGE:image"
# input_stream: "ROI:face_roi"
# input_side_packet: "WITH_ATTENTION:with_attention"
# output_stream: "LANDMARKS:face_landmarks"
# }
@@ -20,8 +25,17 @@ input_stream: "IMAGE:image"
# ROI (region of interest) within the given image where a face is located.
# (NormalizedRect)
input_stream: "ROI:roi"
# Whether to run face mesh model with attention on lips and eyes. (bool)
# Attention provides more accuracy on lips and eye regions as well as iris
# landmarks.
input_side_packet: "WITH_ATTENTION:with_attention"
# 468 face landmarks within the given ROI. (NormalizedLandmarkList)
# 468 or 478 facial landmarks within the given ROI. (NormalizedLandmarkList)
#
# Number of landmarks depends on the WITH_ATTENTION flag. If it's `true` - then
# there will be 478 landmarks with refined lips, eyes and irises (10 extra
# landmarks are for irises), otherwise 468 non-refined landmarks are returned.
#
# NOTE: if a face is not present within the given ROI, for this particular
# timestamp there will not be an output packet in the LANDMARKS stream. However,
# the MediaPipe framework will internally inform the downstream calculators of
@@ -46,31 +60,63 @@ node: {
}
}
# Loads the face landmarks TF Lite model.
node {
calculator: "FaceLandmarksModelLoader"
input_side_packet: "WITH_ATTENTION:with_attention"
output_side_packet: "MODEL:model"
}
# Generates a single side packet containing a TensorFlow Lite op resolver that
# supports custom ops needed by the model used in this graph.
node {
calculator: "TfLiteCustomOpResolverCalculator"
output_side_packet: "op_resolver"
}
# Runs a TensorFlow Lite model on CPU that takes an image tensor and outputs a
# vector of tensors representing, for instance, detection boxes/keypoints and
# scores.
node {
calculator: "InferenceCalculator"
input_stream: "TENSORS:input_tensors"
input_side_packet: "MODEL:model"
input_side_packet: "CUSTOM_OP_RESOLVER:op_resolver"
output_stream: "TENSORS:output_tensors"
options: {
[mediapipe.InferenceCalculatorOptions.ext] {
model_path: "mediapipe/modules/face_landmark/face_landmark.tflite"
delegate { xnnpack {} }
}
}
}
# Splits a vector of tensors into multiple vectors.
# Splits a vector of tensors into landmark tensors and face flag tensor.
node {
calculator: "SplitTensorVectorCalculator"
calculator: "SwitchContainer"
input_side_packet: "ENABLE:with_attention"
input_stream: "output_tensors"
output_stream: "landmark_tensors"
output_stream: "face_flag_tensor"
options: {
[mediapipe.SplitVectorCalculatorOptions.ext] {
ranges: { begin: 0 end: 1 }
ranges: { begin: 1 end: 2 }
[mediapipe.SwitchContainerOptions.ext] {
contained_node: {
calculator: "SplitTensorVectorCalculator"
options: {
[mediapipe.SplitVectorCalculatorOptions.ext] {
ranges: { begin: 0 end: 1 }
ranges: { begin: 1 end: 2 }
}
}
}
contained_node: {
calculator: "SplitTensorVectorCalculator"
options: {
[mediapipe.SplitVectorCalculatorOptions.ext] {
ranges: { begin: 0 end: 6 }
ranges: { begin: 6 end: 7 }
}
}
}
}
}
}
@@ -112,14 +158,18 @@ node {
# Decodes the landmark tensors into a vector of landmarks, where the landmark
# coordinates are normalized by the size of the input image to the model.
node {
calculator: "TensorsToLandmarksCalculator"
calculator: "SwitchContainer"
input_side_packet: "ENABLE:with_attention"
input_stream: "TENSORS:ensured_landmark_tensors"
output_stream: "NORM_LANDMARKS:landmarks"
output_stream: "LANDMARKS:landmarks"
options: {
[mediapipe.TensorsToLandmarksCalculatorOptions.ext] {
num_landmarks: 468
input_image_width: 192
input_image_height: 192
[mediapipe.SwitchContainerOptions.ext] {
contained_node: {
calculator: "TensorsToFaceLandmarks"
}
contained_node: {
calculator: "TensorsToFaceLandmarksWithAttention"
}
}
}
}
mediapipe/modules/face_landmark/face_landmark_front_cpu.pbtxt
+25 -13
View File
@@ -8,13 +8,19 @@
#
# It is required that "face_landmark.tflite" is available at
# "mediapipe/modules/face_landmark/face_landmark.tflite"
# path during execution.
# path during execution if `with_attention` is not set or set to `false`.
#
# It is required that "face_landmark_with_attention.tflite" is available at
# "mediapipe/modules/face_landmark/face_landmark_with_attention.tflite"
# path during execution if `with_attention` is set to `true`.
#
# EXAMPLE:
# node {
# calculator: "FaceLandmarkFrontCpu"
# input_stream: "IMAGE:image"
# input_side_packet: "NUM_FACES:num_faces"
# input_side_packet: "USE_PREV_LANDMARKS:use_prev_landmarks"
# input_side_packet: "WITH_ATTENTION:with_attention"
# output_stream: "LANDMARKS:multi_face_landmarks"
# }
@@ -26,6 +32,15 @@ input_stream: "IMAGE:image"
# Max number of faces to detect/track. (int)
input_side_packet: "NUM_FACES:num_faces"
# Whether landmarks on the previous image should be used to help localize
# landmarks on the current image. (bool)
input_side_packet: "USE_PREV_LANDMARKS:use_prev_landmarks"
# Whether to run face mesh model with attention on lips and eyes. (bool)
# Attention provides more accuracy on lips and eye regions as well as iris
# landmarks.
input_side_packet: "WITH_ATTENTION:with_attention"
# Collection of detected/predicted faces, each represented as a list of 468 face
# landmarks. (std::vector<NormalizedLandmarkList>)
# NOTE: there will not be an output packet in the LANDMARKS stream for this
@@ -44,23 +59,19 @@ output_stream: "ROIS_FROM_LANDMARKS:face_rects_from_landmarks"
# (std::vector<NormalizedRect>)
output_stream: "ROIS_FROM_DETECTIONS:face_rects_from_detections"
# Defines whether landmarks on the previous image should be used to help
# localize landmarks on the current image.
node {
name: "ConstantSidePacketCalculator"
calculator: "ConstantSidePacketCalculator"
output_side_packet: "PACKET:use_prev_landmarks"
options: {
[mediapipe.ConstantSidePacketCalculatorOptions.ext]: {
packet { bool_value: true }
}
}
}
# When the optional input side packet "use_prev_landmarks" is either absent or
# set to true, uses the landmarks on the previous image to help localize
# landmarks on the current image.
node {
calculator: "GateCalculator"
input_side_packet: "ALLOW:use_prev_landmarks"
input_stream: "prev_face_rects_from_landmarks"
output_stream: "gated_prev_face_rects_from_landmarks"
options: {
[mediapipe.GateCalculatorOptions.ext] {
allow: true
}
}
}
# Determines if an input vector of NormalizedRect has a size greater than or
@@ -186,6 +197,7 @@ node {
calculator: "FaceLandmarkCpu"
input_stream: "IMAGE:landmarks_loop_image"
input_stream: "ROI:face_rect"
input_side_packet: "WITH_ATTENTION:with_attention"
output_stream: "LANDMARKS:face_landmarks"
}
mediapipe/modules/face_landmark/face_landmark_front_cpu_image.pbtxt
+31 -3
View File
@@ -8,8 +8,17 @@ input_stream: "IMAGE:image"
# Max number of faces to detect/track. (int)
input_side_packet: "NUM_FACES:num_faces"
# The original input image. (Image)
output_stream: "IMAGE:image"
# Whether landmarks on the previous image should be used to help localize
# landmarks on the current image. (bool)
input_side_packet: "USE_PREV_LANDMARKS:use_prev_landmarks"
# Whether to run face mesh model with attention on lips and eyes. (bool)
# Attention provides more accuracy on lips and eye regions as well as iris
# landmarks.
input_side_packet: "WITH_ATTENTION:with_attention"
# The throttled input image. (Image)
output_stream: "IMAGE:throttled_image"
# Collection of detected/predicted faces, each represented as a list of 468 face
# landmarks. (std::vector<NormalizedLandmarkList>)
# NOTE: there will not be an output packet in the LANDMARKS stream for this
@@ -28,10 +37,27 @@ output_stream: "ROIS_FROM_LANDMARKS:face_rects_from_landmarks"
# (std::vector<NormalizedRect>)
output_stream: "ROIS_FROM_DETECTIONS:face_rects_from_detections"
node {
calculator: "FlowLimiterCalculator"
input_stream: "image"
input_stream: "FINISHED:multi_face_landmarks"
input_stream_info: {
tag_index: "FINISHED"
back_edge: true
}
output_stream: "throttled_image"
options: {
[mediapipe.FlowLimiterCalculatorOptions.ext] {
max_in_flight: 1
max_in_queue: 1
}
}
}
# Converts Image to ImageFrame for FaceLandmarkFrontCpu to consume.
node {
calculator: "FromImageCalculator"
input_stream: "IMAGE:image"
input_stream: "IMAGE:throttled_image"
output_stream: "IMAGE_CPU:raw_image_frame"
output_stream: "SOURCE_ON_GPU:is_gpu_image"
}
@@ -52,6 +78,8 @@ node {
calculator: "FaceLandmarkFrontCpu"
input_stream: "IMAGE:image_frame"
input_side_packet: "NUM_FACES:num_faces"
input_side_packet: "USE_PREV_LANDMARKS:use_prev_landmarks"
input_side_packet: "WITH_ATTENTION:with_attention"
output_stream: "LANDMARKS:multi_face_landmarks"
output_stream: "DETECTIONS:face_detections"
output_stream: "ROIS_FROM_LANDMARKS:face_rects_from_landmarks"
mediapipe/modules/face_landmark/face_landmark_front_gpu.pbtxt
+25 -13
View File
@@ -8,13 +8,19 @@
#
# It is required that "face_landmark.tflite" is available at
# "mediapipe/modules/face_landmark/face_landmark.tflite"
# path during execution.
# path during execution if `with_attention` is not set or set to `false`.
#
# It is required that "face_landmark_with_attention.tflite" is available at
# "mediapipe/modules/face_landmark/face_landmark_with_attention.tflite"
# path during execution if `with_attention` is set to `true`.
#
# EXAMPLE:
# node {
# calculator: "FaceLandmarkFrontGpu"
# input_stream: "IMAGE:image"
# input_side_packet: "NUM_FACES:num_faces"
# input_side_packet: "USE_PREV_LANDMARKS:use_prev_landmarks"
# input_side_packet: "WITH_ATTENTION:with_attention"
# output_stream: "LANDMARKS:multi_face_landmarks"
# }
@@ -26,6 +32,15 @@ input_stream: "IMAGE:image"
# Max number of faces to detect/track. (int)
input_side_packet: "NUM_FACES:num_faces"
# Whether landmarks on the previous image should be used to help localize
# landmarks on the current image. (bool)
input_side_packet: "USE_PREV_LANDMARKS:use_prev_landmarks"
# Whether to run face mesh model with attention on lips and eyes. (bool)
# Attention provides more accuracy on lips and eye regions as well as iris
# landmarks.
input_side_packet: "WITH_ATTENTION:with_attention"
# Collection of detected/predicted faces, each represented as a list of 468 face
# landmarks. (std::vector<NormalizedLandmarkList>)
# NOTE: there will not be an output packet in the LANDMARKS stream for this
@@ -44,23 +59,19 @@ output_stream: "ROIS_FROM_LANDMARKS:face_rects_from_landmarks"
# (std::vector<NormalizedRect>)
output_stream: "ROIS_FROM_DETECTIONS:face_rects_from_detections"
# Defines whether landmarks on the previous image should be used to help
# localize landmarks on the current image.
node {
name: "ConstantSidePacketCalculator"
calculator: "ConstantSidePacketCalculator"
output_side_packet: "PACKET:use_prev_landmarks"
options: {
[mediapipe.ConstantSidePacketCalculatorOptions.ext]: {
packet { bool_value: true }
}
}
}
# When the optional input side packet "use_prev_landmarks" is either absent or
# set to true, uses the landmarks on the previous image to help localize
# landmarks on the current image.
node {
calculator: "GateCalculator"
input_side_packet: "ALLOW:use_prev_landmarks"
input_stream: "prev_face_rects_from_landmarks"
output_stream: "gated_prev_face_rects_from_landmarks"
options: {
[mediapipe.GateCalculatorOptions.ext] {
allow: true
}
}
}
# Determines if an input vector of NormalizedRect has a size greater than or
@@ -186,6 +197,7 @@ node {
calculator: "FaceLandmarkGpu"
input_stream: "IMAGE:landmarks_loop_image"
input_stream: "ROI:face_rect"
input_side_packet: "WITH_ATTENTION:with_attention"
output_stream: "LANDMARKS:face_landmarks"
}
mediapipe/modules/face_landmark/face_landmark_front_gpu_image.pbtxt
+31 -3
View File
@@ -8,8 +8,17 @@ input_stream: "IMAGE:image"
# Max number of faces to detect/track. (int)
input_side_packet: "NUM_FACES:num_faces"
# The original input image. (Image)
output_stream: "IMAGE:image"
# Whether landmarks on the previous image should be used to help localize
# landmarks on the current image. (bool)
input_side_packet: "USE_PREV_LANDMARKS:use_prev_landmarks"
# Whether to run face mesh model with attention on lips and eyes. (bool)
# Attention provides more accuracy on lips and eye regions as well as iris
# landmarks.
input_side_packet: "WITH_ATTENTION:with_attention"
# The throttled input image. (Image)
output_stream: "IMAGE:throttled_image"
# Collection of detected/predicted faces, each represented as a list of 468 face
# landmarks. (std::vector<NormalizedLandmarkList>)
# NOTE: there will not be an output packet in the LANDMARKS stream for this
@@ -28,10 +37,27 @@ output_stream: "ROIS_FROM_LANDMARKS:face_rects_from_landmarks"
# (std::vector<NormalizedRect>)
output_stream: "ROIS_FROM_DETECTIONS:face_rects_from_detections"
node {
calculator: "FlowLimiterCalculator"
input_stream: "image"
input_stream: "FINISHED:multi_face_landmarks"
input_stream_info: {
tag_index: "FINISHED"
back_edge: true
}
output_stream: "throttled_image"
options: {
[mediapipe.FlowLimiterCalculatorOptions.ext] {
max_in_flight: 1
max_in_queue: 1
}
}
}
# Converts Image to GpuBuffer for FaceLandmarkFrontGpu to consume.
node {
calculator: "FromImageCalculator"
input_stream: "IMAGE:image"
input_stream: "IMAGE:throttled_image"
output_stream: "IMAGE_GPU:raw_gpu_buffer"
output_stream: "SOURCE_ON_GPU:is_gpu_image"
}
@@ -52,6 +78,8 @@ node {
calculator: "FaceLandmarkFrontGpu"
input_stream: "IMAGE:gpu_buffer"
input_side_packet: "NUM_FACES:num_faces"
input_side_packet: "USE_PREV_LANDMARKS:use_prev_landmarks"
input_side_packet: "WITH_ATTENTION:with_attention"
output_stream: "LANDMARKS:multi_face_landmarks"
output_stream: "DETECTIONS:face_detections"
output_stream: "ROIS_FROM_LANDMARKS:face_rects_from_landmarks"
mediapipe/modules/face_landmark/face_landmark_gpu.pbtxt
+70 -19
View File
@@ -3,13 +3,18 @@
#
# It is required that "face_landmark.tflite" is available at
# "mediapipe/modules/face_landmark/face_landmark.tflite"
# path during execution.
# path during execution if `with_attention` is not set or set to `false`.
#
# It is required that "face_landmark_with_attention.tflite" is available at
# "mediapipe/modules/face_landmark/face_landmark_with_attention.tflite"
# path during execution if `with_attention` is set to `true`.
#
# EXAMPLE:
# node {
# calculator: "FaceLandmarkGpu"
# input_stream: "IMAGE:image"
# input_stream: "ROI:face_roi"
# input_side_packet: "WITH_ATTENTION:with_attention"
# output_stream: "LANDMARKS:face_landmarks"
# }
@@ -20,8 +25,17 @@ input_stream: "IMAGE:image"
# ROI (region of interest) within the given image where a face is located.
# (NormalizedRect)
input_stream: "ROI:roi"
# Whether to run face mesh model with attention on lips and eyes. (bool)
# Attention provides more accuracy on lips and eye regions as well as iris
# landmarks.
input_side_packet: "WITH_ATTENTION:with_attention"
# 468 face landmarks within the given ROI. (NormalizedLandmarkList)
# 468 or 478 facial landmarks within the given ROI. (NormalizedLandmarkList)
#
# Number of landmarks depends on the WITH_ATTENTION flag. If it's `true` - then
# there will be 478 landmarks with refined lips, eyes and irises (10 extra
# landmarks are for irises), otherwise 468 non-refined landmarks are returned.
#
# NOTE: if a face is not present within the given ROI, for this particular
# timestamp there will not be an output packet in the LANDMARKS stream. However,
# the MediaPipe framework will internally inform the downstream calculators of
@@ -47,30 +61,63 @@ node: {
}
}
# Loads the face landmarks TF Lite model.
node {
calculator: "FaceLandmarksModelLoader"
input_side_packet: "WITH_ATTENTION:with_attention"
output_side_packet: "MODEL:model"
}
# Generates a single side packet containing a TensorFlow Lite op resolver that
# supports custom ops needed by the model used in this graph.
node {
calculator: "TfLiteCustomOpResolverCalculator"
output_side_packet: "op_resolver"
}
# Runs a TensorFlow Lite model on GPU that takes an image tensor and outputs a
# vector of GPU tensors representing, for instance, detection boxes/keypoints
# and scores.
node {
calculator: "InferenceCalculator"
input_stream: "TENSORS:input_tensors"
input_side_packet: "MODEL:model"
input_side_packet: "CUSTOM_OP_RESOLVER:op_resolver"
output_stream: "TENSORS:output_tensors"
options: {
[mediapipe.InferenceCalculatorOptions.ext] {
model_path: "mediapipe/modules/face_landmark/face_landmark.tflite"
# Do not remove. Used for generation of XNNPACK/NNAPI graphs.
}
}
}
# Splits a vector of tensors into multiple vectors.
# Splits a vector of tensors into landmark tensors and face flag tensor.
node {
calculator: "SplitTensorVectorCalculator"
calculator: "SwitchContainer"
input_side_packet: "ENABLE:with_attention"
input_stream: "output_tensors"
output_stream: "landmark_tensors"
output_stream: "face_flag_tensor"
options: {
[mediapipe.SplitVectorCalculatorOptions.ext] {
ranges: { begin: 0 end: 1 }
ranges: { begin: 1 end: 2 }
options {
[mediapipe.SwitchContainerOptions.ext] {
contained_node: {
calculator: "SplitTensorVectorCalculator"
options: {
[mediapipe.SplitVectorCalculatorOptions.ext] {
ranges: { begin: 0 end: 1 }
ranges: { begin: 1 end: 2 }
}
}
}
contained_node: {
calculator: "SplitTensorVectorCalculator"
options: {
[mediapipe.SplitVectorCalculatorOptions.ext] {
ranges: { begin: 0 end: 6 }
ranges: { begin: 6 end: 7 }
}
}
}
}
}
}
@@ -81,11 +128,11 @@ node {
calculator: "TensorsToFloatsCalculator"
input_stream: "TENSORS:face_flag_tensor"
output_stream: "FLOAT:face_presence_score"
options {
[mediapipe.TensorsToFloatsCalculatorOptions.ext] {
activation: SIGMOID
}
options: {
[mediapipe.TensorsToFloatsCalculatorOptions.ext] {
activation: SIGMOID
}
}
}
# Applies a threshold to the confidence score to determine whether a face is
@@ -112,14 +159,18 @@ node {
# Decodes the landmark tensors into a vector of landmarks, where the landmark
# coordinates are normalized by the size of the input image to the model.
node {
calculator: "TensorsToLandmarksCalculator"
calculator: "SwitchContainer"
input_side_packet: "ENABLE:with_attention"
input_stream: "TENSORS:ensured_landmark_tensors"
output_stream: "NORM_LANDMARKS:landmarks"
output_stream: "LANDMARKS:landmarks"
options: {
[mediapipe.TensorsToLandmarksCalculatorOptions.ext] {
num_landmarks: 468
input_image_width: 192
input_image_height: 192
[mediapipe.SwitchContainerOptions.ext] {
contained_node: {
calculator: "TensorsToFaceLandmarks"
}
contained_node: {
calculator: "TensorsToFaceLandmarksWithAttention"
}
}
}
}
mediapipe/modules/face_landmark/face_landmark_with_attention.tflite
Executable
BIN
View File
Binary file not shown.
mediapipe/modules/face_landmark/face_landmarks_model_loader.pbtxt
+58
View File
@@ -0,0 +1,58 @@
# MediaPipe graph to load a selected face landmarks TF Lite model.
type: "FaceLandmarksModelLoader"
# Whether to run face mesh model with attention on lips and eyes. (bool)
# Attention provides more accuracy on lips and eye regions as well as iris
# landmarks.
input_side_packet: "WITH_ATTENTION:with_attention"
# TF Lite model represented as a FlatBuffer.
# (std::unique_ptr<tflite::FlatBufferModel, std::function<void(tflite::FlatBufferModel*)>>)
output_side_packet: "MODEL:model"
# Determines path to the desired face landmark model file based on specification
# in the input side packet.
node {
calculator: "SwitchContainer"
input_side_packet: "ENABLE:with_attention"
output_side_packet: "PACKET:model_path"
options: {
[mediapipe.SwitchContainerOptions.ext] {
contained_node: {
calculator: "ConstantSidePacketCalculator"
options: {
[mediapipe.ConstantSidePacketCalculatorOptions.ext]: {
packet {
string_value: "mediapipe/modules/face_landmark/face_landmark.tflite"
}
}
}
}
contained_node: {
calculator: "ConstantSidePacketCalculator"
options: {
[mediapipe.ConstantSidePacketCalculatorOptions.ext]: {
packet {
string_value: "mediapipe/modules/face_landmark/face_landmark_with_attention.tflite"
}
}
}
}
}
}
}
# Loads the file in the specified path into a blob.
node {
calculator: "LocalFileContentsCalculator"
input_side_packet: "FILE_PATH:model_path"
output_side_packet: "CONTENTS:model_blob"
}
# Converts the input blob into a TF Lite model.
node {
calculator: "TfLiteModelCalculator"
input_side_packet: "MODEL_BLOB:model_blob"
output_side_packet: "MODEL:model"
}
mediapipe/modules/face_landmark/tensors_to_face_landmarks.pbtxt
+24
View File
@@ -0,0 +1,24 @@
# MediaPipe graph to transform single tensor into 468 facial landmarks.
type: "TensorsToFaceLandmarks"
# Vector with a single tensor that contains 468 landmarks. (std::vector<Tensor>)
input_stream: "TENSORS:tensors"
# 468 facial landmarks (NormalizedLandmarkList)
output_stream: "LANDMARKS:landmarks"
# Decodes the landmark tensors into a vector of lanmarks, where the landmark
# coordinates are normalized by the size of the input image to the model.
node {
calculator: "TensorsToLandmarksCalculator"
input_stream: "TENSORS:tensors"
output_stream: "NORM_LANDMARKS:landmarks"
options: {
[mediapipe.TensorsToLandmarksCalculatorOptions.ext] {
num_landmarks: 468
input_image_width: 192
input_image_height: 192
}
}
}
mediapipe/modules/face_landmark/tensors_to_face_landmarks_with_attention.pbtxt
+299
View File
@@ -0,0 +1,299 @@
# MediaPipe graph to transform model output tensors into 478 facial landmarks
# with refined lips, eyes and irises.
type: "TensorsToFaceLandmarksWithAttention"
# Vector with a six tensors to parse landmarks from. (std::vector<Tensor>)
# Landmark tensors order:
# - mesh_tensor
# - lips_tensor
# - left_eye_tensor
# - right_eye_tensor
# - left_iris_tensor
# - right_iris_tensor
input_stream: "TENSORS:tensors"
# 478 facial landmarks (NormalizedLandmarkList)
output_stream: "LANDMARKS:landmarks"
# Splits a vector of tensors into multiple vectors.
node {
calculator: "SplitTensorVectorCalculator"
input_stream: "tensors"
output_stream: "mesh_tensor"
output_stream: "lips_tensor"
output_stream: "left_eye_tensor"
output_stream: "right_eye_tensor"
output_stream: "left_iris_tensor"
output_stream: "right_iris_tensor"
options: {
[mediapipe.SplitVectorCalculatorOptions.ext] {
ranges: { begin: 0 end: 1 }
ranges: { begin: 1 end: 2 }
ranges: { begin: 2 end: 3 }
ranges: { begin: 3 end: 4 }
ranges: { begin: 4 end: 5 }
ranges: { begin: 5 end: 6 }
}
}
}
# Decodes mesh landmarks tensor into a vector of normalized lanmarks.
node {
calculator: "TensorsToLandmarksCalculator"
input_stream: "TENSORS:mesh_tensor"
output_stream: "NORM_LANDMARKS:mesh_landmarks"
options: {
[mediapipe.TensorsToLandmarksCalculatorOptions.ext] {
num_landmarks: 468
input_image_width: 192
input_image_height: 192
}
}
}
# Decodes lips landmarks tensor into a vector of normalized lanmarks.
node {
calculator: "TensorsToLandmarksCalculator"
input_stream: "TENSORS:lips_tensor"
output_stream: "NORM_LANDMARKS:lips_landmarks"
options: {
[mediapipe.TensorsToLandmarksCalculatorOptions.ext] {
num_landmarks: 80
input_image_width: 192
input_image_height: 192
}
}
}
# Decodes left eye landmarks tensor into a vector of normalized lanmarks.
node {
calculator: "TensorsToLandmarksCalculator"
input_stream: "TENSORS:left_eye_tensor"
output_stream: "NORM_LANDMARKS:left_eye_landmarks"
options: {
[mediapipe.TensorsToLandmarksCalculatorOptions.ext] {
num_landmarks: 71
input_image_width: 192
input_image_height: 192
}
}
}
# Decodes right eye landmarks tensor into a vector of normalized lanmarks.
node {
calculator: "TensorsToLandmarksCalculator"
input_stream: "TENSORS:right_eye_tensor"
output_stream: "NORM_LANDMARKS:right_eye_landmarks"
options: {
[mediapipe.TensorsToLandmarksCalculatorOptions.ext] {
num_landmarks: 71
input_image_width: 192
input_image_height: 192
}
}
}
# Decodes left iris landmarks tensor into a vector of normalized lanmarks.
node {
calculator: "TensorsToLandmarksCalculator"
input_stream: "TENSORS:left_iris_tensor"
output_stream: "NORM_LANDMARKS:left_iris_landmarks"
options: {
[mediapipe.TensorsToLandmarksCalculatorOptions.ext] {
num_landmarks: 5
input_image_width: 192
input_image_height: 192
}
}
}
# Decodes right iris landmarks tensor into a vector of normalized lanmarks.
node {
calculator: "TensorsToLandmarksCalculator"
input_stream: "TENSORS:right_iris_tensor"
output_stream: "NORM_LANDMARKS:right_iris_landmarks"
options: {
[mediapipe.TensorsToLandmarksCalculatorOptions.ext] {
num_landmarks: 5
input_image_width: 192
input_image_height: 192
}
}
}
# Refine mesh landmarks with lips, eyes and irises.
node {
calculator: "LandmarksRefinementCalculator"
input_stream: "LANDMARKS:0:mesh_landmarks"
input_stream: "LANDMARKS:1:lips_landmarks"
input_stream: "LANDMARKS:2:left_eye_landmarks"
input_stream: "LANDMARKS:3:right_eye_landmarks"
input_stream: "LANDMARKS:4:left_iris_landmarks"
input_stream: "LANDMARKS:5:right_iris_landmarks"
output_stream: "REFINED_LANDMARKS:landmarks"
options: {
[mediapipe.LandmarksRefinementCalculatorOptions.ext] {
# 0 - mesh
refinement: {
indexes_mapping: [
0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19,
20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36,
37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53,
54, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70,
71, 72, 73, 74, 75, 76, 77, 78, 79, 80, 81, 82, 83, 84, 85, 86, 87,
88, 89, 90, 91, 92, 93, 94, 95, 96, 97, 98, 99, 100, 101, 102, 103,
104, 105, 106, 107, 108, 109, 110, 111, 112, 113, 114, 115, 116, 117,
118, 119, 120, 121, 122, 123, 124, 125, 126, 127, 128, 129, 130, 131,
132, 133, 134, 135, 136, 137, 138, 139, 140, 141, 142, 143, 144, 145,
146, 147, 148, 149, 150, 151, 152, 153, 154, 155, 156, 157, 158, 159,
160, 161, 162, 163, 164, 165, 166, 167, 168, 169, 170, 171, 172, 173,
174, 175, 176, 177, 178, 179, 180, 181, 182, 183, 184, 185, 186, 187,
188, 189, 190, 191, 192, 193, 194, 195, 196, 197, 198, 199, 200, 201,
202, 203, 204, 205, 206, 207, 208, 209, 210, 211, 212, 213, 214, 215,
216, 217, 218, 219, 220, 221, 222, 223, 224, 225, 226, 227, 228, 229,
230, 231, 232, 233, 234, 235, 236, 237, 238, 239, 240, 241, 242, 243,
244, 245, 246, 247, 248, 249, 250, 251, 252, 253, 254, 255, 256, 257,
258, 259, 260, 261, 262, 263, 264, 265, 266, 267, 268, 269, 270, 271,
272, 273, 274, 275, 276, 277, 278, 279, 280, 281, 282, 283, 284, 285,
286, 287, 288, 289, 290, 291, 292, 293, 294, 295, 296, 297, 298, 299,
300, 301, 302, 303, 304, 305, 306, 307, 308, 309, 310, 311, 312, 313,
314, 315, 316, 317, 318, 319, 320, 321, 322, 323, 324, 325, 326, 327,
328, 329, 330, 331, 332, 333, 334, 335, 336, 337, 338, 339, 340, 341,
342, 343, 344, 345, 346, 347, 348, 349, 350, 351, 352, 353, 354, 355,
356, 357, 358, 359, 360, 361, 362, 363, 364, 365, 366, 367, 368, 369,
370, 371, 372, 373, 374, 375, 376, 377, 378, 379, 380, 381, 382, 383,
384, 385, 386, 387, 388, 389, 390, 391, 392, 393, 394, 395, 396, 397,
398, 399, 400, 401, 402, 403, 404, 405, 406, 407, 408, 409, 410, 411,
412, 413, 414, 415, 416, 417, 418, 419, 420, 421, 422, 423, 424, 425,
426, 427, 428, 429, 430, 431, 432, 433, 434, 435, 436, 437, 438, 439,
440, 441, 442, 443, 444, 445, 446, 447, 448, 449, 450, 451, 452, 453,
454, 455, 456, 457, 458, 459, 460, 461, 462, 463, 464, 465, 466, 467
]
z_refinement: { copy {} }
}
# 1 - lips
refinement: {
indexes_mapping: [
# Lower outer.
61, 146, 91, 181, 84, 17, 314, 405, 321, 375, 291,
# Upper outer (excluding corners).
185, 40, 39, 37, 0, 267, 269, 270, 409,
# Lower inner.
78, 95, 88, 178, 87, 14, 317, 402, 318, 324, 308,
# Upper inner (excluding corners).
191, 80, 81, 82, 13, 312, 311, 310, 415,
# Lower semi-outer.
76, 77, 90, 180, 85, 16, 315, 404, 320, 307, 306,
# Upper semi-outer (excluding corners).
184, 74, 73, 72, 11, 302, 303, 304, 408,
# Lower semi-inner.
62, 96, 89, 179, 86, 15, 316, 403, 319, 325, 292,
# Upper semi-inner (excluding corners).
183, 42, 41, 38, 12, 268, 271, 272, 407
]
z_refinement: { none {} }
}
# 2 - left eye
refinement: {
indexes_mapping: [
# Lower contour.
33, 7, 163, 144, 145, 153, 154, 155, 133,
# upper contour (excluding corners).
246, 161, 160, 159, 158, 157, 173,
# Halo x2 lower contour.
130, 25, 110, 24, 23, 22, 26, 112, 243,
# Halo x2 upper contour (excluding corners).
247, 30, 29, 27, 28, 56, 190,
# Halo x3 lower contour.
226, 31, 228, 229, 230, 231, 232, 233, 244,
# Halo x3 upper contour (excluding corners).
113, 225, 224, 223, 222, 221, 189,
# Halo x4 upper contour (no lower because of mesh structure) or
# eyebrow inner contour.
35, 124, 46, 53, 52, 65,
# Halo x5 lower contour.
143, 111, 117, 118, 119, 120, 121, 128, 245,
# Halo x5 upper contour (excluding corners) or eyebrow outer contour.
156, 70, 63, 105, 66, 107, 55, 193
]
z_refinement: { none {} }
}
# 3 - right eye
refinement: {
indexes_mapping: [
# Lower contour.
263, 249, 390, 373, 374, 380, 381, 382, 362,
# Upper contour (excluding corners).
466, 388, 387, 386, 385, 384, 398,
# Halo x2 lower contour.
359, 255, 339, 254, 253, 252, 256, 341, 463,
# Halo x2 upper contour (excluding corners).
467, 260, 259, 257, 258, 286, 414,
# Halo x3 lower contour.
446, 261, 448, 449, 450, 451, 452, 453, 464,
# Halo x3 upper contour (excluding corners).
342, 445, 444, 443, 442, 441, 413,
# Halo x4 upper contour (no lower because of mesh structure) or
# eyebrow inner contour.
265, 353, 276, 283, 282, 295,
# Halo x5 lower contour.
372, 340, 346, 347, 348, 349, 350, 357, 465,
# Halo x5 upper contour (excluding corners) or eyebrow outer contour.
383, 300, 293, 334, 296, 336, 285, 417
]
z_refinement: { none {} }
}
# 4 - left iris
refinement: {
indexes_mapping: [
# Center.
468,
# Iris right edge.
469,
# Iris top edge.
470,
# Iris left edge.
471,
# Iris bottom edge.
472
]
z_refinement: {
assign_average: {
indexes_for_average: [
# Lower contour.
33, 7, 163, 144, 145, 153, 154, 155, 133,
# Upper contour (excluding corners).
246, 161, 160, 159, 158, 157, 173
]
}
}
}
# 5 - right iris
refinement: {
indexes_mapping: [
# Center.
473,
# Iris right edge.
474,
# Iris top edge.
475,
# Iris left edge.
476,
# Iris bottom edge.
477
]
z_refinement: {
assign_average: {
indexes_for_average: [
# Lower contour.
263, 249, 390, 373, 374, 380, 381, 382, 362,
# Upper contour (excluding corners).
466, 388, 387, 386, 385, 384, 398
]
}
}
}
}
}
}
mediapipe/modules/hand_landmark/BUILD
+2
View File
@@ -92,6 +92,7 @@ mediapipe_simple_subgraph(
register_as = "HandLandmarkTrackingCpuImage",
deps = [
":hand_landmark_tracking_cpu",
"//mediapipe/calculators/core:flow_limiter_calculator",
"//mediapipe/calculators/image:image_transformation_calculator",
"//mediapipe/calculators/util:from_image_calculator",
],
@@ -103,6 +104,7 @@ mediapipe_simple_subgraph(
register_as = "HandLandmarkTrackingGpuImage",
deps = [
":hand_landmark_tracking_gpu",
"//mediapipe/calculators/core:flow_limiter_calculator",
"//mediapipe/calculators/image:image_transformation_calculator",
"//mediapipe/calculators/util:from_image_calculator",
],
mediapipe/modules/hand_landmark/hand_landmark_sparse.tflite
BIN
View File
Binary file not shown.
mediapipe/modules/hand_landmark/hand_landmark_tracking_cpu.pbtxt
+12 -12
View File
@@ -14,6 +14,10 @@ input_stream: "IMAGE:image"
# Max number of hands to detect/track. (int)
input_side_packet: "NUM_HANDS:num_hands"
# Whether landmarks on the previous image should be used to help localize
# landmarks on the current image. (bool)
input_side_packet: "USE_PREV_LANDMARKS:use_prev_landmarks"
# Collection of detected/predicted hands, each represented as a list of
# landmarks. (std::vector<NormalizedLandmarkList>)
# NOTE: there will not be an output packet in the LANDMARKS stream for this
@@ -38,23 +42,19 @@ output_stream: "HAND_ROIS_FROM_LANDMARKS:hand_rects"
# (std::vector<NormalizedRect>)
output_stream: "HAND_ROIS_FROM_PALM_DETECTIONS:hand_rects_from_palm_detections"
# Defines whether landmarks on the previous image should be used to help
# localize landmarks on the current image.
node {
name: "ConstantSidePacketCalculator"
calculator: "ConstantSidePacketCalculator"
output_side_packet: "PACKET:use_prev_landmarks"
options: {
[mediapipe.ConstantSidePacketCalculatorOptions.ext]: {
packet { bool_value: true }
}
}
}
# When the optional input side packet "use_prev_landmarks" is either absent or
# set to true, uses the landmarks on the previous image to help localize
# landmarks on the current image.
node {
calculator: "GateCalculator"
input_side_packet: "ALLOW:use_prev_landmarks"
input_stream: "prev_hand_rects_from_landmarks"
output_stream: "gated_prev_hand_rects_from_landmarks"
options: {
[mediapipe.GateCalculatorOptions.ext] {
allow: true
}
}
}
# Determines if an input vector of NormalizedRect has a size greater than or
mediapipe/modules/hand_landmark/hand_landmark_tracking_cpu_image.pbtxt
+25 -3
View File
@@ -14,8 +14,12 @@ input_stream: "IMAGE:image"
# Max number of hands to detect/track. (int)
input_side_packet: "NUM_HANDS:num_hands"
# The original input image. (Image)
output_stream: "IMAGE:image"
# Whether landmarks on the previous image should be used to help localize
# landmarks on the current image. (bool)
input_side_packet: "USE_PREV_LANDMARKS:use_prev_landmarks"
# The throttled input image. (Image)
output_stream: "IMAGE:throttled_image"
# Collection of detected/predicted hands, each represented as a list of
# landmarks. (std::vector<NormalizedLandmarkList>)
# NOTE: there will not be an output packet in the LANDMARKS stream for this
@@ -40,10 +44,27 @@ output_stream: "HAND_ROIS_FROM_LANDMARKS:hand_rects"
# (std::vector<NormalizedRect>)
output_stream: "HAND_ROIS_FROM_PALM_DETECTIONS:hand_rects_from_palm_detections"
node {
calculator: "FlowLimiterCalculator"
input_stream: "image"
input_stream: "FINISHED:multi_hand_landmarks"
input_stream_info: {
tag_index: "FINISHED"
back_edge: true
}
output_stream: "throttled_image"
options: {
[mediapipe.FlowLimiterCalculatorOptions.ext] {
max_in_flight: 1
max_in_queue: 1
}
}
}
# Converts Image to ImageFrame for HandLandmarkTrackingCpu to consume.
node {
calculator: "FromImageCalculator"
input_stream: "IMAGE:image"
input_stream: "IMAGE:throttled_image"
output_stream: "IMAGE_CPU:raw_image_frame"
output_stream: "SOURCE_ON_GPU:is_gpu_image"
}
@@ -64,6 +85,7 @@ node {
calculator: "HandLandmarkTrackingCpu"
input_stream: "IMAGE:image_frame"
input_side_packet: "NUM_HANDS:num_hands"
input_side_packet: "USE_PREV_LANDMARKS:use_prev_landmarks"
output_stream: "LANDMARKS:multi_hand_landmarks"
output_stream: "HANDEDNESS:multi_handedness"
output_stream: "PALM_DETECTIONS:palm_detections"
mediapipe/modules/hand_landmark/hand_landmark_tracking_gpu.pbtxt
+12 -12
View File
@@ -14,6 +14,10 @@ input_stream: "IMAGE:image"
# Max number of hands to detect/track. (int)
input_side_packet: "NUM_HANDS:num_hands"
# Whether landmarks on the previous image should be used to help localize
# landmarks on the current image. (bool)
input_side_packet: "USE_PREV_LANDMARKS:use_prev_landmarks"
# Collection of detected/predicted hands, each represented as a list of
# landmarks. (std::vector<NormalizedLandmarkList>)
# NOTE: there will not be an output packet in the LANDMARKS stream for this
@@ -38,23 +42,19 @@ output_stream: "HAND_ROIS_FROM_LANDMARKS:hand_rects"
# (std::vector<NormalizedRect>)
output_stream: "HAND_ROIS_FROM_PALM_DETECTIONS:hand_rects_from_palm_detections"
# Defines whether landmarks on the previous image should be used to help
# localize landmarks on the current image.
node {
name: "ConstantSidePacketCalculator"
calculator: "ConstantSidePacketCalculator"
output_side_packet: "PACKET:use_prev_landmarks"
options: {
[mediapipe.ConstantSidePacketCalculatorOptions.ext]: {
packet { bool_value: true }
}
}
}
# When the optional input side packet "use_prev_landmarks" is either absent or
# set to true, uses the landmarks on the previous image to help localize
# landmarks on the current image.
node {
calculator: "GateCalculator"
input_side_packet: "ALLOW:use_prev_landmarks"
input_stream: "prev_hand_rects_from_landmarks"
output_stream: "gated_prev_hand_rects_from_landmarks"
options: {
[mediapipe.GateCalculatorOptions.ext] {
allow: true
}
}
}
# Determines if an input vector of NormalizedRect has a size greater than or
mediapipe/modules/hand_landmark/hand_landmark_tracking_gpu_image.pbtxt
+25 -3
View File
@@ -14,6 +14,10 @@ input_stream: "IMAGE:image"
# Max number of hands to detect/track. (int)
input_side_packet: "NUM_HANDS:num_hands"
# Whether landmarks on the previous image should be used to help localize
# landmarks on the current image. (bool)
input_side_packet: "USE_PREV_LANDMARKS:use_prev_landmarks"
# Collection of detected/predicted hands, each represented as a list of
# landmarks. (std::vector<NormalizedLandmarkList>)
# NOTE: there will not be an output packet in the LANDMARKS stream for this
@@ -28,8 +32,8 @@ output_stream: "LANDMARKS:multi_hand_landmarks"
# horizontally.
output_stream: "HANDEDNESS:multi_handedness"
# The original input image. (Image)
output_stream: "IMAGE:image"
# The throttled input image. (Image)
output_stream: "IMAGE:throttled_image"
# Extra outputs (for debugging, for instance).
# Detected palms. (std::vector<Detection>)
output_stream: "PALM_DETECTIONS:palm_detections"
@@ -40,10 +44,27 @@ output_stream: "HAND_ROIS_FROM_LANDMARKS:hand_rects"
# (std::vector<NormalizedRect>)
output_stream: "HAND_ROIS_FROM_PALM_DETECTIONS:hand_rects_from_palm_detections"
node {
calculator: "FlowLimiterCalculator"
input_stream: "image"
input_stream: "FINISHED:multi_hand_landmarks"
input_stream_info: {
tag_index: "FINISHED"
back_edge: true
}
output_stream: "throttled_image"
options: {
[mediapipe.FlowLimiterCalculatorOptions.ext] {
max_in_flight: 1
max_in_queue: 1
}
}
}
# Converts Image to GpuBuffer for HandLandmarkTrackingGpu to consume.
node {
calculator: "FromImageCalculator"
input_stream: "IMAGE:image"
input_stream: "IMAGE:throttled_image"
output_stream: "IMAGE_GPU:raw_gpu_buffer"
output_stream: "SOURCE_ON_GPU:is_gpu_image"
}
@@ -64,6 +85,7 @@ node {
calculator: "HandLandmarkTrackingGpu"
input_stream: "IMAGE:gpu_buffer"
input_side_packet: "NUM_HANDS:num_hands"
input_side_packet: "USE_PREV_LANDMARKS:use_prev_landmarks"
output_stream: "LANDMARKS:multi_hand_landmarks"
output_stream: "HANDEDNESS:multi_handedness"
output_stream: "PALM_DETECTIONS:palm_detections"
mediapipe/modules/holistic_landmark/holistic_landmark_cpu.pbtxt
+10
View File
@@ -35,6 +35,7 @@
# input_side_packet: "SMOOTH_LANDMARKS:smooth_landmarks"
# input_side_packet: "ENABLE_SEGMENTATION:enable_segmentation"
# input_side_packet: "SMOOTH_SEGMENTATION:smooth_segmentation"
# input_side_packet: "USE_PREV_LANDMARKS:use_prev_landmarks"
# output_stream: "POSE_LANDMARKS:pose_landmarks"
# output_stream: "FACE_LANDMARKS:face_landmarks"
# output_stream: "LEFT_HAND_LANDMARKS:left_hand_landmarks"
@@ -69,6 +70,10 @@ input_side_packet: "ENABLE_SEGMENTATION:enable_segmentation"
# jitter. If unspecified, functions as set to true. (bool)
input_side_packet: "SMOOTH_SEGMENTATION:smooth_segmentation"
# Whether landmarks on the previous image should be used to help localize
# landmarks on the current image. (bool)
input_side_packet: "USE_PREV_LANDMARKS:use_prev_landmarks"
# Pose landmarks. (NormalizedLandmarkList)
# 33 pose landmarks.
output_stream: "POSE_LANDMARKS:pose_landmarks"
@@ -81,6 +86,9 @@ output_stream: "RIGHT_HAND_LANDMARKS:right_hand_landmarks"
# 468 face landmarks. (NormalizedLandmarkList)
output_stream: "FACE_LANDMARKS:face_landmarks"
# Segmentation mask. (ImageFrame in ImageFormat::VEC32F1)
output_stream: "SEGMENTATION_MASK:segmentation_mask"
# Debug outputs
output_stream: "POSE_ROI:pose_landmarks_roi"
output_stream: "POSE_DETECTION:pose_detection"
@@ -93,8 +101,10 @@ node {
input_side_packet: "SMOOTH_LANDMARKS:smooth_landmarks"
input_side_packet: "ENABLE_SEGMENTATION:enable_segmentation"
input_side_packet: "SMOOTH_SEGMENTATION:smooth_segmentation"
input_side_packet: "USE_PREV_LANDMARKS:use_prev_landmarks"
output_stream: "LANDMARKS:pose_landmarks"
output_stream: "WORLD_LANDMARKS:pose_world_landmarks"
output_stream: "SEGMENTATION_MASK:segmentation_mask"
output_stream: "ROI_FROM_LANDMARKS:pose_landmarks_roi"
output_stream: "DETECTION:pose_detection"
}
mediapipe/modules/holistic_landmark/holistic_landmark_gpu.pbtxt
+10
View File
@@ -35,6 +35,7 @@
# input_side_packet: "SMOOTH_LANDMARKS:smooth_landmarks"
# input_side_packet: "ENABLE_SEGMENTATION:enable_segmentation"
# input_side_packet: "SMOOTH_SEGMENTATION:smooth_segmentation"
# input_side_packet: "USE_PREV_LANDMARKS:use_prev_landmarks"
# output_stream: "POSE_LANDMARKS:pose_landmarks"
# output_stream: "FACE_LANDMARKS:face_landmarks"
# output_stream: "LEFT_HAND_LANDMARKS:left_hand_landmarks"
@@ -69,6 +70,10 @@ input_side_packet: "ENABLE_SEGMENTATION:enable_segmentation"
# jitter. If unspecified, functions as set to true. (bool)
input_side_packet: "SMOOTH_SEGMENTATION:smooth_segmentation"
# Whether landmarks on the previous image should be used to help localize
# landmarks on the current image. (bool)
input_side_packet: "USE_PREV_LANDMARKS:use_prev_landmarks"
# Pose landmarks. (NormalizedLandmarkList)
# 33 pose landmarks.
output_stream: "POSE_LANDMARKS:pose_landmarks"
@@ -81,6 +86,9 @@ output_stream: "RIGHT_HAND_LANDMARKS:right_hand_landmarks"
# 468 face landmarks. (NormalizedLandmarkList)
output_stream: "FACE_LANDMARKS:face_landmarks"
# Segmentation mask. (GpuBuffer in RGBA, with the same mask values in R and A)
output_stream: "SEGMENTATION_MASK:segmentation_mask"
# Debug outputs
output_stream: "POSE_ROI:pose_landmarks_roi"
output_stream: "POSE_DETECTION:pose_detection"
@@ -93,8 +101,10 @@ node {
input_side_packet: "SMOOTH_LANDMARKS:smooth_landmarks"
input_side_packet: "ENABLE_SEGMENTATION:enable_segmentation"
input_side_packet: "SMOOTH_SEGMENTATION:smooth_segmentation"
input_side_packet: "USE_PREV_LANDMARKS:use_prev_landmarks"
output_stream: "LANDMARKS:pose_landmarks"
output_stream: "WORLD_LANDMARKS:pose_world_landmarks"
output_stream: "SEGMENTATION_MASK:segmentation_mask"
output_stream: "ROI_FROM_LANDMARKS:pose_landmarks_roi"
output_stream: "DETECTION:pose_detection"
}
mediapipe/modules/objectron/objectron_cpu.pbtxt
+11 -13
View File
@@ -9,6 +9,9 @@ input_side_packet: "MODEL_PATH:box_landmark_model_path"
input_side_packet: "LABELS_CSV:allowed_labels"
# Max number of objects to detect/track. (int)
input_side_packet: "MAX_NUM_OBJECTS:max_num_objects"
# Whether landmarks on the previous image should be used to help localize
# landmarks on the current image. (bool)
input_side_packet: "USE_PREV_LANDMARKS:use_prev_landmarks"
# Bounding box landmarks topology definition.
# The numbers are indices in the box_landmarks list.
#
@@ -48,24 +51,19 @@ node {
output_side_packet: "MODEL:box_landmark_model"
}
# Defines whether landmarks from the previous video frame should be used to help
# predict landmarks on the current video frame.
node {
name: "ConstantSidePacketCalculator"
calculator: "ConstantSidePacketCalculator"
output_side_packet: "PACKET:use_prev_landmarks"
options: {
[mediapipe.ConstantSidePacketCalculatorOptions.ext]: {
packet { bool_value: true }
}
}
}
# When the optional input side packet "use_prev_landmarks" is either absent or
# set to true, uses the landmarks on the previous image to help localize
# landmarks on the current image.
node {
calculator: "GateCalculator"
input_side_packet: "ALLOW:use_prev_landmarks"
input_stream: "prev_box_rects_from_landmarks"
output_stream: "gated_prev_box_rects_from_landmarks"
options: {
[mediapipe.GateCalculatorOptions.ext] {
allow: true
}
}
}
# Determines if an input vector of NormalizedRect has a size greater than or
mediapipe/modules/objectron/objectron_gpu.pbtxt
+11 -13
View File
@@ -8,28 +8,26 @@ input_stream: "IMAGE_GPU:image"
input_side_packet: "LABELS_CSV:allowed_labels"
# Max number of objects to detect/track. (int)
input_side_packet: "MAX_NUM_OBJECTS:max_num_objects"
# Whether landmarks on the previous image should be used to help localize
# landmarks on the current image. (bool)
input_side_packet: "USE_PREV_LANDMARKS:use_prev_landmarks"
# Collection of detected 3D objects, represented as a FrameAnnotation.
output_stream: "FRAME_ANNOTATION:detected_objects"
# Defines whether landmarks from the previous video frame should be used to help
# predict landmarks on the current video frame.
node {
name: "ConstantSidePacketCalculator"
calculator: "ConstantSidePacketCalculator"
output_side_packet: "PACKET:use_prev_landmarks"
options: {
[mediapipe.ConstantSidePacketCalculatorOptions.ext]: {
packet { bool_value: true }
}
}
}
# When the optional input side packet "use_prev_landmarks" is either absent or
# set to true, uses the landmarks on the previous image to help localize
# landmarks on the current image.
node {
calculator: "GateCalculator"
input_side_packet: "ALLOW:use_prev_landmarks"
input_stream: "prev_box_rects_from_landmarks"
output_stream: "gated_prev_box_rects_from_landmarks"
options: {
[mediapipe.GateCalculatorOptions.ext] {
allow: true
}
}
}
# Determines if an input vector of NormalizedRect has a size greater than or
mediapipe/modules/pose_landmark/pose_landmark_cpu.pbtxt
+13 -12
View File
@@ -21,6 +21,7 @@
# input_side_packet: "SMOOTH_LANDMARKS:smooth_landmarks"
# input_side_packet: "ENABLE_SEGMENTATION:enable_segmentation"
# input_side_packet: "SMOOTH_SEGMENTATION:smooth_segmentation"
# input_side_packet: "USE_PREV_LANDMARKS:use_prev_landmarks"
# input_stream: "IMAGE:image"
# output_stream: "LANDMARKS:pose_landmarks"
# output_stream: "SEGMENTATION_MASK:segmentation_mask"
@@ -48,6 +49,10 @@ input_side_packet: "SMOOTH_SEGMENTATION:smooth_segmentation"
# functions as set to 1. (int)
input_side_packet: "MODEL_COMPLEXITY:model_complexity"
# Whether landmarks on the previous image should be used to help localize
# landmarks on the current image. (bool)
input_side_packet: "USE_PREV_LANDMARKS:use_prev_landmarks"
# Pose landmarks. (NormalizedLandmarkList)
# We have 33 landmarks (see pose_landmark_topology.svg), and there are other
# auxiliary key points.
@@ -110,23 +115,19 @@ output_stream: "ROI_FROM_LANDMARKS:pose_rect_from_landmarks"
# Regions of interest calculated based on pose detections. (NormalizedRect)
output_stream: "ROI_FROM_DETECTION:pose_rect_from_detection"
# Defines whether landmarks on the previous image should be used to help
# localize landmarks on the current image.
node {
name: "ConstantSidePacketCalculator"
calculator: "ConstantSidePacketCalculator"
output_side_packet: "PACKET:use_prev_landmarks"
options: {
[mediapipe.ConstantSidePacketCalculatorOptions.ext]: {
packet { bool_value: true }
}
}
}
# When the optional input side packet "use_prev_landmarks" is either absent or
# set to true, uses the landmarks on the previous image to help localize
# landmarks on the current image.
node {
calculator: "GateCalculator"
input_side_packet: "ALLOW:use_prev_landmarks"
input_stream: "prev_pose_rect_from_landmarks"
output_stream: "gated_prev_pose_rect_from_landmarks"
options: {
[mediapipe.GateCalculatorOptions.ext] {
allow: true
}
}
}
# Checks if there's previous pose rect calculated from landmarks.
mediapipe/modules/pose_landmark/pose_landmark_gpu.pbtxt
+13 -12
View File
@@ -21,6 +21,7 @@
# input_side_packet: "SMOOTH_LANDMARKS:smooth_landmarks"
# input_side_packet: "ENABLE_SEGMENTATION:enable_segmentation"
# input_side_packet: "SMOOTH_SEGMENTATION:smooth_segmentation"
# input_side_packet: "USE_PREV_LANDMARKS:use_prev_landmarks"
# input_stream: "IMAGE:image"
# output_stream: "LANDMARKS:pose_landmarks"
# output_stream: "SEGMENTATION_MASK:segmentation_mask"
@@ -48,6 +49,10 @@ input_side_packet: "SMOOTH_SEGMENTATION:smooth_segmentation"
# functions as set to 1. (int)
input_side_packet: "MODEL_COMPLEXITY:model_complexity"
# Whether landmarks on the previous image should be used to help localize
# landmarks on the current image. (bool)
input_side_packet: "USE_PREV_LANDMARKS:use_prev_landmarks"
# Pose landmarks. (NormalizedLandmarkList)
# We have 33 landmarks (see pose_landmark_topology.svg), and there are other
# auxiliary key points.
@@ -110,23 +115,19 @@ output_stream: "ROI_FROM_LANDMARKS:pose_rect_from_landmarks"
# Regions of interest calculated based on pose detections. (NormalizedRect)
output_stream: "ROI_FROM_DETECTION:pose_rect_from_detection"
# Defines whether landmarks on the previous image should be used to help
# localize landmarks on the current image.
node {
name: "ConstantSidePacketCalculator"
calculator: "ConstantSidePacketCalculator"
output_side_packet: "PACKET:use_prev_landmarks"
options: {
[mediapipe.ConstantSidePacketCalculatorOptions.ext]: {
packet { bool_value: true }
}
}
}
# When the optional input side packet "use_prev_landmarks" is either absent or
# set to true, uses the landmarks on the previous image to help localize
# landmarks on the current image.
node {
calculator: "GateCalculator"
input_side_packet: "ALLOW:use_prev_landmarks"
input_stream: "prev_pose_rect_from_landmarks"
output_stream: "gated_prev_pose_rect_from_landmarks"
options: {
[mediapipe.GateCalculatorOptions.ext] {
allow: true
}
}
}
# Checks if there's previous pose rect calculated from landmarks.
mediapipe/modules/selfie_segmentation/BUILD
+2
View File
@@ -71,6 +71,7 @@ mediapipe_simple_subgraph(
register_as = "SelfieSegmentationCpuImage",
deps = [
":selfie_segmentation_cpu",
"//mediapipe/calculators/core:flow_limiter_calculator",
"//mediapipe/calculators/image:image_transformation_calculator",
"//mediapipe/calculators/util:from_image_calculator",
"//mediapipe/calculators/util:to_image_calculator",
@@ -83,6 +84,7 @@ mediapipe_simple_subgraph(
register_as = "SelfieSegmentationGpuImage",
deps = [
":selfie_segmentation_gpu",
"//mediapipe/calculators/core:flow_limiter_calculator",
"//mediapipe/calculators/image:image_transformation_calculator",
"//mediapipe/calculators/util:from_image_calculator",
"//mediapipe/calculators/util:to_image_calculator",
mediapipe/modules/selfie_segmentation/selfie_segmentation_cpu_image.pbtxt
+20 -3
View File
@@ -5,8 +5,8 @@ type: "SelfieSegmentationCpuImage"
# Input image. (Image)
input_stream: "IMAGE:image"
# The original input image. (Image)
output_stream: "IMAGE:image"
# The throttled input image. (Image)
output_stream: "IMAGE:throttled_image"
# An integer 0 or 1. Use 0 to select a general-purpose model (operating on a
# 256x256 tensor), and 1 to select a model (operating on a 256x144 tensor) more
@@ -16,10 +16,27 @@ input_side_packet: "MODEL_SELECTION:model_selection"
# Segmentation mask. (Image)
output_stream: "SEGMENTATION_MASK:segmentation_mask"
node {
calculator: "FlowLimiterCalculator"
input_stream: "image"
input_stream: "FINISHED:segmentation_mask"
input_stream_info: {
tag_index: "FINISHED"
back_edge: true
}
output_stream: "throttled_image"
options: {
[mediapipe.FlowLimiterCalculatorOptions.ext] {
max_in_flight: 1
max_in_queue: 1
}
}
}
# Converts Image to ImageFrame for SelfieSegmentationCpu to consume.
node {
calculator: "FromImageCalculator"
input_stream: "IMAGE:image"
input_stream: "IMAGE:throttled_image"
output_stream: "IMAGE_CPU:raw_image_frame"
output_stream: "SOURCE_ON_GPU:is_gpu_image"
}
mediapipe/modules/selfie_segmentation/selfie_segmentation_gpu_image.pbtxt
+20 -3
View File
@@ -5,8 +5,8 @@ type: "SelfieSegmentationGpuImage"
# Input image. (Image)
input_stream: "IMAGE:image"
# The original input image. (Image)
output_stream: "IMAGE:image"
# The throttled input image. (Image)
output_stream: "IMAGE:throttled_image"
# An integer 0 or 1. Use 0 to select a general-purpose model (operating on a
# 256x256 tensor), and 1 to select a model (operating on a 256x144 tensor) more
@@ -16,10 +16,27 @@ input_side_packet: "MODEL_SELECTION:model_selection"
# Segmentation mask. (Image)
output_stream: "SEGMENTATION_MASK:segmentation_mask"
node {
calculator: "FlowLimiterCalculator"
input_stream: "image"
input_stream: "FINISHED:segmentation_mask"
input_stream_info: {
tag_index: "FINISHED"
back_edge: true
}
output_stream: "throttled_image"
options: {
[mediapipe.FlowLimiterCalculatorOptions.ext] {
max_in_flight: 1
max_in_queue: 1
}
}
}
# Converts Image to ImageFrame for SelfieSegmentationGpu to consume.
node {
calculator: "FromImageCalculator"
input_stream: "IMAGE:image"
input_stream: "IMAGE:throttled_image"
output_stream: "IMAGE_GPU:raw_gpu_buffer"
output_stream: "SOURCE_ON_GPU:is_gpu_image"
}