Update README.md

README.md

@@ -42,7 +42,7 @@ inference with optional 4-flip test-time augmentation. At evaluation time, water
 reconstructed from the binary prediction using the JRC prior, enabling 3-class IoU reporting
 (no-flood / flood / waterbody) against the original 3-class ground truth.
 
-- **Developed by:** AISEHack Theme 1 Team (ANRF AISEHack Phase 2)
+- **Developed by:** AI-Hackers
 - **Funded by:** ANRF (Anusandhan National Research Foundation) AISEHack Programme
 - **Model type:** Semantic segmentation — geospatial / Earth observation
 - **Input modalities:** Optical (HLS-style 6-band) + SAR (HH, HV, SAR_diff, log_HH, log_HV) + JRC binary water prior
@@ -52,11 +52,8 @@ reconstructed from the binary prediction using the JRC prior, enabling 3-class I
 
 ### Model Sources
 
-- **Repository:** *(GitHub link — add after public release)*
+
 - **Model weights:** `prithvi_best-v7.ckpt` — epoch 125, val flood IoU 0.368 *(Google Drive / Zenodo link)*
-- **Demo:** FloodSense web app *(Vercel link — add after deployment)*
-- **Training code:** `best.py` in repository root
-- **Evaluation code:** `evaluate_model.py`, `quick_eval.py`
 
 ---