large model inference result not correct?

[tianleiwu]

I tried it in Windows:

python -m samexporter.inference ^
    --encoder_model output_models\sam2_hiera_large.encoder.onnx  ^
    --decoder_model output_models\sam2_hiera_large.decoder.onnx  ^
    --image images\truck.jpg  ^
    --prompt images\truck_prompt.json  ^
    --output output_images\sam2_truck.png  ^
    --sam_variant sam2  ^
    --show

Snippet of output is like the following:

[vietanhdev]

v0.5.0 fixes shared SAM2 prompt validation, per-label mask inputs, candidate scoring, original-size scaling, and pins the current official source. SAM2.1 Tiny now matches PyTorch at >0.995 IoU on retained point/box cases. I have not claimed the original Large example is resolved without its exact prompt/output model, so this remains open for verification. Please retest with v0.5.0 and attach the prompt JSON if Large still differs.