tessera saliency models
The ONNX models that tessera, a file tagging tool, downloads on first use for saliency detection and smart thumbnail cropping. Both are exports of Ultralytics YOLOv8 nano models with a 640×640 input.
| File | Model | Size | SHA-256 |
|---|---|---|---|
yolov8n-pose.onnx |
YOLOv8n-pose: people and 17 body keypoints | 13,514,518 bytes | 6e9dd5921b13b9d1bc41effe46ce6e6b33ea342cd501ed7481c6ffe148958eb6 |
yolov8n.onnx |
YOLOv8n: 80 COCO object classes | 12,851,087 bytes | b0ada942ffcbc3be0efac651d60fc15ed7cb9ec56554525fccc3b95aa906460c |
tessera downloads them from a fixed revision of this repository and checks these hashes before it uses a file.
License
YOLOv8 is developed by Ultralytics and released under the GNU Affero General Public License v3.0, as is tessera.