Create README.md

README.md

@@ -0,0 +1,72 @@
+---
+license: mit
+base_model: CIDAS/clipseg-rd64-refined
+tags:
+- image-segmentation
+- semantic-segmentation
+- computer-vision
+- crack-detection
+- infrastructure
+- clipseg
+datasets:
+- roboflow
+metrics:
+- iou
+- dice
+---
+
+# CrackSeg
+
+Fine-tuned [CLIPSeg](https://huggingface.co/CIDAS/clipseg-rd64-refined) for pixel-wise surface crack detection. Given an image of any surface, the model returns a binary segmentation mask highlighting crack regions.
+
+## Model Performance
+
+| Metric | Score |
+|--------|-------|
+| Dice Score | 0.612 |
+| mIoU | 0.716 |
+
+## Live Demo
+
+Try it on [HuggingFace Spaces](https://huggingface.co/spaces/primus29/crackseg).
+
+## Training Details
+
+- **Dataset:** 14,000+ crack images (Roboflow, COCO format)
+- **Fine-tuning:** Partial — decoder fully unfrozen + last 2 layers of CLIP vision encoder + last 1 layer of CLIP text encoder
+- **Loss:** Focal Loss (α=0.75, γ=2.0)
+- **Optimizer:** AdamW with differential learning rates
+- **Scheduler:** CosineAnnealingLR
+- **Early stopping:** patience = 5
+
+## Usage
+
+```python
+import torch
+from huggingface_hub import hf_hub_download
+from transformers import AutoProcessor, CLIPSegForImageSegmentation
+from PIL import Image
+
+processor = AutoProcessor.from_pretrained("CIDAS/clipseg-rd64-refined")
+model = CLIPSegForImageSegmentation.from_pretrained("CIDAS/clipseg-rd64-refined")
+
+path = hf_hub_download(repo_id="primus29/crackseg", filename="best_model.pth")
+checkpoint = torch.load(path, map_location="cpu", weights_only=False)
+model.load_state_dict(checkpoint['model_state_dict'])
+model.eval()
+
+image = Image.open("your_image.jpg")
+inputs = processor(text="segment crack", images=image, return_tensors="pt", padding=True)
+
+with torch.no_grad():
+    outputs = model(**inputs)
+
+mask = torch.sigmoid(outputs.logits).squeeze()
+mask = (mask > 0.5).float()
+```
+
+## Limitations
+
+- Shadow regions can be misidentified as cracks
+- Performance degrades on very thin hairline cracks
+- Trained primarily on surface/concrete crack data; may not generalize to all materials