anudit commited on
Commit
b8d15a7
·
verified ·
1 Parent(s): 146ae88

Add 1024x1024 variant for small inputs

Browse files
Files changed (2) hide show
  1. README.md +3 -0
  2. onnx/model_fp16_1024.onnx +3 -0
README.md CHANGED
@@ -21,6 +21,9 @@ ONNX export of [ZhengPeng7/BiRefNet_lite-2K](https://huggingface.co/ZhengPeng7/B
21
  |---|---|---|---|---|
22
  | `onnx/model_fp16.onnx` | fp16 weights, fp32 I/O | `input_image` `[1,3,1440,2560]` (ImageNet-normalised RGB) | `output_image` `[1,1,1440,2560]` logits (apply sigmoid) | ~99 MB |
23
 
 
 
 
24
  The input size is **fixed**: resize (stretch, no aspect preservation — this matches training) your image
25
  to 2560×1440, run, then resize the mask back.
26
 
 
21
  |---|---|---|---|---|
22
  | `onnx/model_fp16.onnx` | fp16 weights, fp32 I/O | `input_image` `[1,3,1440,2560]` (ImageNet-normalised RGB) | `output_image` `[1,1,1440,2560]` logits (apply sigmoid) | ~99 MB |
23
 
24
+ `onnx/model_fp16_1024.onnx` is the same weights exported at a fixed `[1,3,1024,1024]` input (~93 MB, ~4× faster, ~¼ the memory);
25
+ `birefnet-webgpu` uses it automatically for small source images.
26
+
27
  The input size is **fixed**: resize (stretch, no aspect preservation — this matches training) your image
28
  to 2560×1440, run, then resize the mask back.
29
 
onnx/model_fp16_1024.onnx ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ab11fddbecb0bbb8b07aa698d07f8fe15771b2714c992404a90588d7df2dc028
3
+ size 93445455