anime-character-detector
February 13, 2026 · View on GitHub
YOLOX-based anime character detector running in browser with ONNX Runtime Web (WebGPU/WASM)
Links
- Web Demo: https://ksasao.github.io/anime-character-detector/
- Model Download: docs/character.onnx
- Python Code: python/
Model Information
| Property | Value |
|---|---|
| Base Model | YOLOX-s |
| Input Size | 640×640 |
| Classes | Single class: "character" |
| mAP@50 | 82.4% |
| mAP@50-95 | 57.7% |
| Training Images | 3,253 |
| Annotations | 15,901 |
Features
- Client-side Processing: All inference runs in your browser using ONNX Runtime Web
- Multiple Input Methods: Drag & drop, file selection, or paste from clipboard
- WebGPU/WASM Support: Hardware acceleration when available
- HEIC/HEIF Support: Automatic conversion to PNG for iOS images
- Adjustable Parameters: Score threshold and NMS IoU threshold controls
License
Apache License 2.0 - see LICENSE for details.