# MediaPipe graph that performs multi-hand tracking with TensorFlow Lite on GPU.
# Used in the examples in
# mediapipe/examples/android/src/java/com/mediapipe/apps/handtrackinggpu.

# GPU image. (GpuBuffer)
input_stream: "input_video"

# Max number of hands to detect/process. (int)
input_side_packet: "num_hands"

# Model complexity (0 or 1). (int)
input_side_packet: "model_complexity"

# GPU image. (GpuBuffer)
output_stream: "output_video"
# Collection of detected/predicted hands, each represented as a list of
# landmarks. (std::vector<NormalizedLandmarkList>)
output_stream: "hand_landmarks"

# Throttles the images flowing downstream for flow control. It passes through
# the very first incoming image unaltered, and waits for downstream nodes
# (calculators and subgraphs) in the graph to finish their tasks before it
# passes through another image. All images that come in while waiting are
# dropped, limiting the number of in-flight images in most part of the graph to
# 1. This prevents the downstream nodes from queuing up incoming images and data
# excessively, which leads to increased latency and memory usage, unwanted in
# real-time mobile applications. It also eliminates unnecessarily computation,
# e.g., the output produced by a node may get dropped downstream if the
# subsequent nodes are still busy processing previous inputs.
node {
  calculator: "FlowLimiterCalculator"
  input_stream: "input_video"
  input_stream: "FINISHED:output_video"
  input_stream_info: {
    tag_index: "FINISHED"
    back_edge: true
  }
  output_stream: "throttled_input_video"
}

# Detects/tracks hand landmarks.
node {
  calculator: "HandLandmarkTrackingGpu"
  input_stream: "IMAGE:throttled_input_video"
  input_side_packet: "MODEL_COMPLEXITY:model_complexity"
  input_side_packet: "NUM_HANDS:num_hands"
  output_stream: "LANDMARKS:hand_landmarks"
  output_stream: "HANDEDNESS:handedness"
  output_stream: "PALM_DETECTIONS:palm_detections"
  output_stream: "HAND_ROIS_FROM_LANDMARKS:hand_rects_from_landmarks"
  output_stream: "HAND_ROIS_FROM_PALM_DETECTIONS:hand_rects_from_palm_detections"
}

# Subgraph that renders annotations and overlays them on top of the input
# images (see hand_renderer_gpu.pbtxt).
node {
  calculator: "HandRendererSubgraph"
  input_stream: "IMAGE:throttled_input_video"
  input_stream: "DETECTIONS:palm_detections"
  input_stream: "LANDMARKS:hand_landmarks"
  input_stream: "HANDEDNESS:handedness"
  input_stream: "NORM_RECTS:0:hand_rects_from_palm_detections"
  input_stream: "NORM_RECTS:1:hand_rects_from_landmarks"
  output_stream: "IMAGE:output_video"
}