Initial commit.

app.py

@@ -172,7 +172,46 @@ def predict(input_img):
 
 title = "Drive&Segment"
 description = 'Gradio Demo accompanying paper "Drive&Segment: Unsupervised Semantic Segmentation of Urban Scenes via Cross-modal Distillation"\nBecause of the CPU-only inference, it might take up to 20s for large images.\nRight now, it uses the Segmenter model trained on nuScenes and with a simplified inference scheme (for the sake of speed).'
-article = "<p style='text-align: center'><a href='https://vobecant.github.io/DriveAndSegment/' target='_blank'>Project Page</a> | <a href='https://github.com/vobecant/DriveAndSegment' target='_blank'>Github</a></p>"
+# article = "<p style='text-align: center'><a href='https://vobecant.github.io/DriveAndSegment/' target='_blank'>Project Page</a> | <a href='https://github.com/vobecant/DriveAndSegment' target='_blank'>Github</a></p>"
+article="""
+<h1 align="center">🚙📷 Drive&Segment: Unsupervised Semantic Segmentation of Urban Scenes via Cross-modal Distillation</h1>
+
+<h2 align="center">
+  <a href="https://vobecant.github.io/DriveAndSegment">project page</a> |
+  <a href="#">arXiv</a> |
+  <a href="https://huggingface.co/spaces/vobecant/DaS">Gradio</a> |
+  <a href="https://colab.research.google.com/drive/126tBVYbt1s0STyv8DKhmLoHKpvWcv33H?usp=sharing">Colab</a> |
+  <a href="https://www.youtube.com/watch?v=B9LK-Fxu7ao">video</a> 
+</h2>
+
+## 💫 Highlights
+
+- 🚫🔬 **Unsupervised semantic segmentation:** Drive&Segments proposes learning semantic segmentation in urban scenes without any manual annotation, just from
+the raw non-curated data collected by cars which, equipped with 📷 cameras and 💥 LiDAR sensors.
+- 📷💥 **Multi-modal training:** During the train time our method takes 📷 images and 💥 LiDAR scans as an input, and
+  learns a semantic segmentation model *without using manual annotations*.
+- 📷 **Image-only inference:** During the inference time, Drive&Segments takes *only images* as an input.
+- 🏆 **State-of-the-art performance:** Our best single model based on Segmenter architecture achieves **21.8%** in mIoU on
+  Cityscapes (without any fine-tuning).
+
+![teaser](https://drive.google.com/uc?export=view&id=1MkQmAfBPUomJDUikLhM_Wk8VUNekPb91)
+
+## 📺 Examples
+
+### **Pseudo** segmentation.
+
+Example of **pseudo** segmentation.
+
+![](https://github.com/vobecant/DriveAndSegment/blob/main/sources/video128_blend03_v2_10fps_640px_lanczos.gif)
+
+### Cityscapes segmentation.
+
+Two examples of pseudo segmentation mapped to the 19 ground-truth classes of the Cityscapes dataset by using Hungarian
+algorithm.
+
+![](https://github.com/vobecant/DriveAndSegment/blob/main/sources/video_stuttgart00_remap_blended03_20fps_crop.gif)
+![](https://github.com/vobecant/DriveAndSegment/blob/main/sources/video_stuttgart01_remap_blended03_20fps_crop2.gif)
+"""
 examples = [ #'examples/img5.jpeg',
             'examples/100.jpeg',
             'examples/39076.jpeg',