support mask prompt for video tracking

grounded_sam2_tracking_demo_with_gd1.5.py

@@ -129,18 +129,34 @@ elif masks.ndim == 4:
 Step 3: Register each object's positive points to video predictor with seperate add_new_points call
 """
 
-# sample the positive points from mask for each objects
-all_sample_points = sample_points_from_masks(masks=masks, num_points=10)
+PROMPT_TYPE_FOR_VIDEO = "mask" # or "point"
 
-for object_id, (label, points) in enumerate(zip(OBJECTS, all_sample_points), start=1):
-    labels = np.ones((points.shape[0]), dtype=np.int32)
-    _, out_obj_ids, out_mask_logits = video_predictor.add_new_points(
-        inference_state=inference_state,
-        frame_idx=ann_frame_idx,
-        obj_id=object_id,
-        points=points,
-        labels=labels,
-    )
+assert PROMPT_TYPE_FOR_VIDEO in ["point", "mask"]
+
+# If you are using point prompts, we uniformly sample positive points based on the mask
+if PROMPT_TYPE_FOR_VIDEO == "point":
+    # sample the positive points from mask for each objects
+    all_sample_points = sample_points_from_masks(masks=masks, num_points=10)
+
+    for object_id, (label, points) in enumerate(zip(OBJECTS, all_sample_points), start=1):
+        labels = np.ones((points.shape[0]), dtype=np.int32)
+        _, out_obj_ids, out_mask_logits = video_predictor.add_new_points(
+            inference_state=inference_state,
+            frame_idx=ann_frame_idx,
+            obj_id=object_id,
+            points=points,
+            labels=labels,
+        )
+# Using mask prompt is a more straightforward way
+elif PROMPT_TYPE_FOR_VIDEO == "mask":
+    for object_id, (label, mask) in enumerate(zip(OBJECTS, masks), start=1):
+        labels = np.ones((1), dtype=np.int32)
+        _, out_obj_ids, out_mask_logits = video_predictor.add_new_mask(
+            inference_state=inference_state,
+            frame_idx=ann_frame_idx,
+            obj_id=object_id,
+            mask=mask
+        )
 
 
 """