I build tools for computer vision.
- labelme ★16k+ - label images for segmentation, detection, and classification
- imgviz ★268+ - visualize images and labels without OpenCV or matplotlib
- octomap-python ★100+ - Python bindings for the OctoMap 3D mapping library
- osam ★110+ - run SAM1/2/3, EfficientSAM, YOLO-World, and other promptable vision models locally
- sam3-onnx ★120+ - ONNX export and inference for SAM3
- yolo-world-onnx - ONNX models for YOLO-World open-vocabulary detection
- labelme-satellite-image-demo - annotate satellite imagery with labelme and convert to GeoJSON for QGIS
- gdown ★5.4k+ - download Google Drive files that wget and curl choke on
- video-cli ★136+ - crop, trim, retime, and convert videos with one-line commands
- moviepy-cli - edit videos from the command line via MoviePy
- jqk - render JSON with jq patterns
- imshow - display images from Python with a customizable viewer
- git-hunk - non-interactive git hunk staging for AI agents
- gruff (wip) - lint Python code for maintainability alongside Ruff
- ihq (wip) - externalize git-ignored files to a synced, identity-derived store
- acron (wip) - schedule unattended coding agents on your own server via cron
- pytorch-fcn ★1.7k+ - Fully Convolutional Networks in PyTorch
- pytorch-for-numpy-users ★706+ - PyTorch reference for NumPy users
- morefusion ★238+ - 6D pose estimation from volumetric fusion (CVPR 2020)
- fcn ★215+ - Fully Convolutional Networks in Chainer
- gshell ★113+ - navigate Google Drive as you do on a shell
- reorientbot ★58+ - learning object reorientation for posed placement (ICRA 2022)
- safepicking ★56+ - safe object extraction via object-level mapping (ICRA 2022)