9 projects
online-emotion-detection
Streaming, frame-by-frame facial emotion recognition (HSEmotion) with a unified torch/torchscript/onnx/trt runtime and export-once caching for edge devices.
online-face-detection
Streaming, frame-by-frame face detection (RetinaFace) with a unified torch/torchscript/onnx/trt runtime and export-once caching for edge devices.
gallery-track-lib
A modular video tracking + gallery ReID toolkit built on top of det-v1 detection outputs. This package is the second stage in the Vision Pipeline.
detect-lib
A modular video object detection toolkit with a clean det-v1 JSON schema, pluggable backends, and optional model export.
video-annote
video-annote is a lightweight multi-video annotation tool (PyQt5) for labeling time ranges (“labels”) while reviewing one or more videos.
gaze-estimation-lib
Gaze augmentation stage: attach gaze (yaw/pitch + gaze_vec) using face boxes in upstream JSON.
detect-face-lib
detect-face-lib is a modular face detection + JSON augmentation toolkit that attaches face detections to det-v1(detections) or track-v1(tracked) payloads produced by earlier stages in the Vision Pipeline.
transcribe-audio
A CLI + Python package for automatic speech recognition (ASR) on audio or video using pluggable backends (currently WhisperX).
denoise-audio
Denoise WAV audio using RNNoise, DeepFilterNet, and FacebookResearch Denoiser.
