Instructions to use occurra/object_detection with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- TensorRT
How to use occurra/object_detection with TensorRT:
# No code snippets available yet for this library. # To use this model, check the repository files and the library's documentation. # Want to help? PRs adding snippets are welcome at: # https://github.com/huggingface/huggingface.js
- Notebooks
- Google Colab
- Kaggle
Commit History
14 days ago
-
Export the NMS once for the whole batch
Verified signatureThe Ultralytics NMS of the previous exports was unrolled into one subgraph per image of the batch, which left TensorRT a third idle. These models end in one NMS for the whole batch: EfficientNMS_TRT in FP8 and INT8 (TensorRT EP only), standard ONNX NMS in FP16. Same output (N, 300, 6). Built by agents/python/object_detection/scripts/main.py of occurra/vms #222.49d7b29 -
VMS-1082: Rebuild the NMS models for a batch of 32
Verified signatureThe embedded NMS of these Ultralytics exports answered only the first image of a dynamic batch (export batch defaulted to 1). Rebuilt with batch=32 by agents/python/object_detection/scripts/main.py; every batch size 1..32 checked on TensorRT at every position.58b22d3
-
Initial upload: YOLOv11 ONNX (nano + small × {apple,fp16,fp8,int8} × {640x640,640x480})
Verified signature503f5cd -
initial commit
Verified signaturec67b7b3