display metric names left to first column

app.py

@@ -113,6 +113,10 @@ layout_parts = []
 video_paths = []
 row_heights = [""]*num_rows
 gt_viz_indices = []
+if len(model_viz) > 1:
+    left_pad = font_size*max([len(viz['display_name']) for viz in model_viz[1:]]) + padding*2
+else:
+    left_pad = 0
 for row_idx in range(num_rows):
     for col_idx in range(num_cols):
         vid_idx = len(video_paths)
@@ -120,36 +124,43 @@ for row_idx in range(num_rows):
         if cur_model['name'] == "gt":
             if row_idx < len(gt_viz):
                 video_path = get_path_for_viz(data_base_path, selected_dataset, selected_sequence, cur_model, gt_viz[row_idx])
-                # if not os.path.isfile(video_path):
-                #     raise ValueError("Video path does not exist: " + video_path)
+                if not os.path.isfile(video_path):
+                    raise ValueError("Video path does not exist: " + video_path)
                 gt_viz_indices.append(vid_idx)
             else:
                 continue
         else:
             if row_idx < len(model_viz):
                 video_path = get_path_for_viz(data_base_path, selected_dataset, selected_sequence, cur_model, model_viz[row_idx])
-                # if not os.path.isfile(video_path):
-                #     raise ValueError("Video path does not exist: " + video_path)
+                if not os.path.isfile(video_path):
+                    raise ValueError("Video path does not exist: " + video_path)
             else:
                 continue
 
         if row_heights[row_idx] == "":
             row_heights[row_idx] = "h{}".format(vid_idx)
         if row_idx == 0:
-            # font_size
-            # pad=240:180+23:0:180:white
             pad_height = font_size+padding*2
             pad_txt_str = ",pad={}:{}:0:{}:white".format(w+padding*2, h+font_size+padding*4, pad_height)
             text_str = get_text_str(pad_height, w, cur_model['display_name'], font_path, font_size)
             pad_txt_str = pad_txt_str + "," + text_str
+        elif row_idx > 0 and col_idx == 0:
+            pad_txt_str = ",pad={}:ih:{}:0:white".format(w + left_pad + padding*2, left_pad)
+            text_str = get_text_str("h", left_pad, model_viz[row_idx]['display_name'], font_path, font_size)
+            pad_txt_str = pad_txt_str + "," + text_str
         else:
             pad_txt_str = ""
+
         input_filter_part = "[{}:v]scale={}:-1,{}{}[v{}]".format(vid_idx, w, pad_str, pad_txt_str, vid_idx)
         input_filter_parts.append(input_filter_part)
         xstack_input_part = "[v{}]".format(vid_idx)
         xstack_input_parts.append(xstack_input_part)
         video_paths.append(video_path)
-        layout_w = "+".join(["w{}".format(i) for i in range(col_idx)]) if col_idx > 0 else "0"
+        if row_idx == 0 or col_idx > 0:
+            layout_w_parts = [str(left_pad)] + ["w{}".format(i) for i in range(col_idx)]
+            layout_w = "+".join(layout_w_parts)
+        else:
+            layout_w = "+".join(["w{}".format(i) for i in range(col_idx)]) if col_idx > 0 else "0"
         if cur_model['name'] == "gt":
             layout_h = "+".join(["h{}".format(i) for i in gt_viz_indices[:-1]]) if row_idx > 0 else "0"
         else: