|
|
--- |
|
|
license: mit |
|
|
--- |
|
|
# ONNX models for MobileSAM |
|
|
|
|
|
* Original repository: [MobileSAM (Github)](https://github.com/ChaoningZhang/MobileSAM) |
|
|
* Original weights: [MobileSam (HuggingFace)](https://huggingface.co/dhkim2810/MobileSAM) |
|
|
|
|
|
## Models |
|
|
|
|
|
| Model | Description | |
|
|
|-|-| |
|
|
| mobile_sam_image_encoder.onnx | Image encoder created from MobileSAM weights | |
|
|
| sam_mask_decoder_single.onnx | Mask decoder for SegmentAnything, returns a single mask | |
|
|
| sam_mask_decoder_multi.onnx | Mask decoder for SegmentAnything, returns 4 masks | |
|
|
|
|
|
## Export |
|
|
|
|
|
The image encoder was exported with `export_image_enoder.py` found in this repository. |
|
|
The mask decoders were exported with `export_models.py` from the MobileSAM repository. |
|
|
|
