Gsmv's picture
|
download
raw
871 Bytes

Segment Anything (browser build)

Unmodified ONNX weights from Xenova/slimsam-77-uniform revision 5850ab45f587c112167512ffef949107115e26a0 (Apache-2.0, see LICENSE). That export is SlimSAM-77-uniform (nielsr/slimsam-77-uniform), a compressed Segment Anything model. Segment Anything itself is Apache-2.0 (Meta).

Same pair WaveSpeed's Segment Anything tool runs:

  • vision_encoder.onnx — 23,276,014 bytes. Input pixel_values float32 [1,3,1024,1024]. Outputs image_embeddings and image_positional_embeddings, each [1,256,64,64].
  • prompt_encoder_mask_decoder.onnx — 16,557,892 bytes. Inputs input_points float32 [1,1,N,2], input_labels int64 [1,1,N], plus the two encoder outputs. Outputs iou_scores [1,1,3] and pred_masks [1,1,3,256,256].

Xet Storage Details

Size:
871 Bytes
·
Xet hash:
3561c4625960fe5de3d92c099f26dcf2e9b9dcf47962e6b35aa33a785dc4318e

Xet efficiently stores files, intelligently splitting them into unique chunks and accelerating uploads and downloads. More info.