mli0603 / stereo-transformer

Revisiting Stereo Depth Estimation From a Sequence-to-Sequence Perspective with Transformers. (ICCV 2021 Oral)
Apache License 2.0
664 stars 106 forks source link

The default parameters of the model #95

Closed Sissi-lu closed 4 months ago

Sissi-lu commented 7 months ago

Hi, Could you print the default parameters of the model when using the sttr-light pretrained model (sttr_light_sceneflow_pretrained_model.pth.tar)?

When I did the inference with sttr-light pretrained model, I found the model size mismatch.

Thanks in advance.

Sissi-lu commented 4 months ago

The other branch is the sttr-light model.