Anyuhhh/sign-language-recognition
0
Sign Language Recognition (AutoGluon, local ZIP)
This Space loads an AutoGluon MultiModalPredictor from the ZIP in this repo: autogluon_image_predictor_dir.zip.
Notes
- The model was trained with an older AutoGluon config. The
requirements.txtpins versions (AGMM 0.8.2, PL <2.2, OmegaConf <2.4, Torch 2.1.2) so the checkpoint structure matches and avoids theomegaconf ... get_child('optimization')error. - Keep these files in the repo root:
app.py, the ZIP,requirements.txt(this file). - If you later retrain with a newer AutoGluon, you can relax the pins.
