Update README.md
Browse files
README.md
CHANGED
|
@@ -1,3 +1,12 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
# Dual-View SLaVA-CXR
|
| 2 |
|
| 3 |
**Dual-View SLaVA-CXR** is a vision-language model for structured radiology report generation from frontal and lateral chest X-rays. Built on the Re³ (Recognize–Reason–Report) paradigm and extending the original SLaVA-CXR model, this project integrates dual-view vision fusion and leverages CLIP, BiomedCLIP, and Phi-2 for enhanced anatomical reasoning.
|
|
@@ -118,4 +127,4 @@ Use `llava_phi/generation.ipynb` with both frontal and lateral views, plus a pro
|
|
| 118 |
|
| 119 |
## 📜 License
|
| 120 |
|
| 121 |
-
This repository is provided for academic research purposes only.
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: mit
|
| 3 |
+
metrics:
|
| 4 |
+
- bertscore
|
| 5 |
+
- bleu
|
| 6 |
+
- rouge
|
| 7 |
+
base_model:
|
| 8 |
+
- microsoft/phi-2
|
| 9 |
+
---
|
| 10 |
# Dual-View SLaVA-CXR
|
| 11 |
|
| 12 |
**Dual-View SLaVA-CXR** is a vision-language model for structured radiology report generation from frontal and lateral chest X-rays. Built on the Re³ (Recognize–Reason–Report) paradigm and extending the original SLaVA-CXR model, this project integrates dual-view vision fusion and leverages CLIP, BiomedCLIP, and Phi-2 for enhanced anatomical reasoning.
|
|
|
|
| 127 |
|
| 128 |
## 📜 License
|
| 129 |
|
| 130 |
+
This repository is provided for academic research purposes only.
|