SENet-154 monocular depth (Hu et al. 2018 architecture) — NYU-Depth V2

Trained 8 epochs on every 3. frame, batch 8, AMP. Encoder: timm legacy_senet154 (stem pool0 included in E_senet).

model_senet.pth — final clean state_dict (no module. prefix)
checkpoint_last.pth — resume checkpoint (model + optimizer + scaler + history)

Load with the matching architecture from the training notebook.

Downloads last month: -; Downloads are not tracked for this model. How to track

Inference Providers NEW

This model isn't deployed by any Inference Provider. 🙋 Ask for provider support