Add webcam

app.py

@@ -49,7 +49,16 @@ def dictionary_to_tensor(landmarks_dict: dict) -> torch.Tensor:
     return torch.from_numpy(output)
 
 
-def greet(video):
+def greet(label, video0, video1):
+
+    if label == "Webcam":
+        video = video0
+
+    elif label == "Video upload":
+        video = video1
+
+    else:
+        return {}
 
     data = obtain_pose_data(video)
 
@@ -60,7 +69,6 @@ def greet(video):
         depth_map[:, index, 1] = data.data_hub[identifier + "_Y"]
 
     depth_map = torch.from_numpy(np.copy(depth_map))
-    print(depth_map.shape)
 
     depth_map = tensor_to_dictionary(depth_map)
 
@@ -88,13 +96,14 @@ def greet(video):
     results = torch.nn.functional.softmax(outputs, dim=2).detach().numpy()[0, 0]
 
     results = {GLOSS[i]: float(results[i]) for i in range(100)}
-    print(outputs.shape)
-    print(results)
 
     return results
 
 
 label = gr.outputs.Label(num_top_classes=5)
-demo = gr.Interface(fn=greet, inputs="playable_video", outputs=label)
+demo = gr.Interface(fn=greet, inputs=[gr.Dropdown(["Webcam", "Video"], label="Please select the input type:"), gr.Video(source="webcam", label="Webcam recording"), gr.Video(label="Video upload")], outputs=label,
+                    title="SPOTER Sign language recognition",
+                    description="Lightweight and accurate sign langage recognitoon right in your browser.",
+                    article="This is joint work of Matyas Bohacek and Zhuo Cao. For more info, visit [our website.](www.signlanguagerecognition.com)")
 
 demo.launch()