A set of tools and examples for converting and utilizing powerful vision models, DINOv3 and EdgeTAM (SAM2), within the ONNX ecosystem.
-
Updated
Sep 28, 2026 - Jupyter Notebook
GeezerVisonThe World Wide Web, restored to 800 pixels of glory. |
| Home|About|Gaming cabinet|Archives|How it works|Starter sites |
GeezerVison is showing github.com — Build software better, together
#
edgetamHere are 11 public repositories matching this topic...Language: All
Filter by language
Sort: Most stars
Sort options
Real-time SAM2 segmentation on edge devices - 40x faster C++ inference with ONNX Runtime for iOS/Android deployment
computer-vision tensorrt embedded-ai edge-ai edge-deployment onnx-runtime cpp-inference zero-shot-segmentation ios-deployment sam2 arm-optimization lightweight-ai on-device-inference edgetam real-time-segmentation npu-acceleration mobile-segmentation picosam android-edge prompt-segmentation
Meta's EdgeTAM (SAM 2 for the edge) as four Core ML packages: click once, track the object through video at ~48 ms a frame on the Apple Neural Engine.
macos ios computer-vision object-tracking coreml coremltools video-object-segmentation apple-silicon segment-anything apple-neural-engine sam2 edgetam
Free open-source local AI character swap for video on Windows and Linux. Wan 2.2 Animate on your own NVIDIA GPU through ComfyUI, prep in Rust and ONNX Runtime (EdgeTAM, RT-DETRv2, BiRefNet). No cloud, no account, no per-video cost. A Viggle alternative with an MCP server. Unofficial.
windows linux rust offline mcp cuda nvidia tauri onnxruntime character-swap generative-ai ai-video video-to-video comfyui local-ai sam2 edgetam wan-animate wan-2-2 viggle-alternative
Live, multi-camera object tracking with SAM 2 and EdgeTAM, on one GPU or many. One model per GPU serves every camera: faster than the original code, about 4x less GPU memory, identical masks.
computer-vision gpu inference pytorch batching object-tracking video-segmentation model-serving segment-anything sam2 edgetam
A Flutter mobile application that performs interactive, on-device image segmentation using EdgeTAM (SAM2) encoder/decoder exported to ONNX
mobile-app sam flutter flutter-app segmentation-masks onnxruntime onnx-models segment-anything sam2 edgetam
Click once, track the object through video: EdgeTAM and SAM 2.1 tiny in Rust on ONNX Runtime's CUDA provider. ~42 ms a frame, mean mask IoU 0.97–0.998 vs PyTorch, no Python at run time. Windows and Linux.
windows linux rust computer-vision gpu cuda nvidia ort object-tracking video-segmentation onnx interactive-segmentation video-object-segmentation edge-ai onnxruntime segment-anything sam2 segment-anything-2 sam-2 edgetam
A Flutter Edge AI application demonstrating on-device, zero-shot multimodal object segmentation and real-time tracking. Combines Gemma 4 w/ EdgeTAM & DINOv3
mobile-app segmentation ffi-bindings object-tracking zero-shot zero-shot-learning jni-bindings flutter-app edge-ai segment-anything dinov3 edgetam dinov3-onnx gemma-4
Dither a video — or just one person in it. EdgeTAM tracking, 14 error-diffusion kernels and 18 palettes, entirely in the browser. Optional local accelerator for 2x.
A Flutter application that leverages the Segment Anything Model and ARCore raw depth data to interactively segment objects and calculate their precise real-world dimensions.
Bounded, observable streaming video segmentation built on SAM 2 and EdgeTAM.
Add this topic to your repoTo associate your repository with the edgetam topic, visit your repo's landing page and select "manage topics." |
| You are visitor 01402520 — GeezerVison does not store the pages you look up. Best viewed with Netscape Navigator 4.7 or Internet Explorer 6 at 800×600. No cookies required. |