Views
No views yet
CIDAS/clipseg-rd64-refined backbone, decoder + FiLM
unfrozen) for prompted binary segmentation on drywall imagery. Single
checkpoint covers two classes selected by text prompt:"segment crack" → wall-crack mask"segment taping area" → drywall taping-seam mask| Task | Dice | mIoU | Precision | Recall |
|---|---|---|---|---|
| Crack | 0.672 | 0.531 | — | — |
| Taping | 0.727 | 0.587 | — | — |
1from huggingface_hub import hf_hub_download
2import torch
3from src.models.clipseg_wrapper import CLIPSegFT
4
5ckpt_path = hf_hub_download(repo_id="ravindrakapse/drywall-clipseg", filename="best.pt")
6model = CLIPSegFT(pretrained="CIDAS/clipseg-rd64-refined").cuda()
7state = torch.load(ckpt_path, map_location="cuda")
8model.load_state_dict(state["model"])
9model.eval()load_models.py for the full inference pipeline (letterbox + TTA + un-letterbox).