unimatch GMFlow-scale2-regrefine6-mixdata (optical flow)

Mirror of the gmflow-scale2-regrefine6-mixdata-train320x576-4e7b215d.pth optical-flow checkpoint from autonomousvision/unimatch, re-hosted for use with AmalgamationAI's OpticalFlow component model.

unimatch ships no official HuggingFace repo and distributes weights via a release URL only; this repo exists so the weights resolve through the standard HF model cache.

File: gmflow-scale2-regrefine6-mixdata-train320x576-4e7b215d.pth (sha256 prefix 4e7b215d)
Architecture: GMFlow, num_scales=2, upsample_factor=4, reg_refine=True, num_transformer_layers=6 (load with strict=True).
Upstream source: https://s3.eu-central-1.amazonaws.com/avg-projects/unimatch/pretrained/
License: MIT (Copyright (c) 2022 autonomousvision).
Citation: Xu et al., "Unifying Flow, Stereo and Depth Estimation", TPAMI 2023.

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