Add more md info

app.py

@@ -106,22 +106,13 @@ def greet(label, video0, video1):
 label = gr.outputs.Label(num_top_classes=5, label="Top class probabilities")
 demo = gr.Interface(fn=greet, inputs=[gr.Dropdown(["Webcam", "Video"], label="Please select the input type:", type="value"), gr.Video(source="webcam", label="Webcam recording", type="mp4"), gr.Video(source="upload", label="Video upload", type="mp4")], outputs=label,
                     title="🤟 SPOTER Sign language recognition",
-                    description="""
-                    
-                    Try out our recent model for sign language recognition right in your browser! The model below takes a video of a single sign in the American Sign Language at the input and provides you with probabilities of the lemmas (equivalent to words in natural language).
-                    
-                    ### Our work at CVPR
-                    
-                    Our efforts on lightweight and efficient models for sign language recognition were first introduced at WACV with our SPOTER paper. We now presented a work-in-progress follow-up here at CVPR's AVA workshop. Be sure to check our work and code below:
-                    
-                    - **WACV2022** - Original SPOTER paper - [Paper](), [Code]()
-                    - **CVPR2022 AVA Worshop** - Follow-up WIP – [Extended Abstract](), [Poster]()
-                    
-                    ### How to sign?
-                    
-                    The model wrapped in this demo was trained on [WLASL100](https://dxli94.github.io/WLASL/), so it only knows selected ASL vocabulary. Take a look at these tutorial video examples, try to replicate them yourself, and have them recognized using the webcam capture below. Have fun!
-                    
-                    """,
+                    description="""Try out our recent model for sign language recognition right in your browser! The model below takes a video of a single sign in the American Sign Language at the input and provides you with probabilities of the lemmas (equivalent to words in natural language).
+### Our work at CVPR
+Our efforts on lightweight and efficient models for sign language recognition were first introduced at WACV with our SPOTER paper. We now presented a work-in-progress follow-up here at CVPR's AVA workshop. Be sure to check our work and code below:                
+- **WACV2022** - Original SPOTER paper - [Paper](), [Code]()
+- **CVPR2022 AVA Worshop** - Follow-up WIP – [Extended Abstract](), [Poster]()                 
+### How to sign?       
+The model wrapped in this demo was trained on [WLASL100](https://dxli94.github.io/WLASL/), so it only knows selected ASL vocabulary. Take a look at these tutorial video examples, try to replicate them yourself, and have them recognized using the webcam capture below. Have fun!""",
                     article="This is joint work of [Matyas Bohacek](https://scholar.google.cz/citations?user=wDy1xBwAAAAJ) and [Zhuo Cao](https://www.linkedin.com/in/zhuo-cao-b0787a1aa/?originalSubdomain=hk). For more info, visit [our website.](https://www.signlanguagerecognition.com)",
                     css="""
                             @font-face {