vid: support manual button

src/main.py

@@ -354,11 +354,9 @@ def process_video_rectangle_changed(api: sly.Api, event: sly.Event.Tools.Rectang
         direction = -1 if idx < event_frame_idx else 1
         frame_id = f"{event.video_id}_{frame_index}"
 
-        _diff = get_increment_by_diff(
+        _drow, _dcol = get_increment_by_diff(
             api, event, frame_id, direction, reverse_direction=True, modality="video"
         )
-        # if _diff is None:
-        _drow, _dcol = _diff
 
         drow += _drow * direction
         dcol += _dcol * direction
@@ -464,72 +462,99 @@ def rectangle_changed(api: sly.Api, event: sly.Event.Tools.Rectangle.FigureChang
 
 def process_video(api: sly.Api, event: sly.Event.FigureCreated):
     project_meta = g.get_project_meta(api, event.project_id)
-    figure = api.image.figure.get_info_by_id(event.figure_state["id"])
-    obj_class = project_meta.get_obj_class_by_id(event.figure_state["classId"])
 
     ann_json = api.video.annotation.download(event.video_id)
     key_id_map = sly.KeyIdMap()
     video_ann = sly.VideoAnnotation.from_json(ann_json, project_meta, key_id_map)
     video_info = api.video.get_info_by_id(event.video_id)
 
+    processed_vidfigures: List[sly.VideoFigure] = []
+
     end_frame = min(event.frame + 2, video_info.frames_count)
     start_frame = max(event.frame - 2, 0)
 
     frame_indexes = [x for x in range(end_frame + 1)][start_frame:]
     event_frame_idx = frame_indexes.index(event.frame)
     frame_indexes[:event_frame_idx] = frame_indexes[:event_frame_idx][::-1]
 
-    video_object = sly.VideoObject(
-        obj_class,
-        key=key_id_map.get_object_key(event.object_id),
-        class_id=event.tool_class_id,
-    )
-    video_figure = sly.VideoFigure(
-        video_object,
-        figure.bbox,
-        event_frame_idx,
-        # key=key_id_map.get_object_key(event.object_id),
-        # class_id=event.tool_class_id,
-        track_id=event.object_id,
-    )
+    try:
+        figure = api.image.figure.get_info_by_id(event.figure_state["id"])
+        obj_class = project_meta.get_obj_class_by_id(event.figure_state["classId"])
+        video_object = sly.VideoObject(
+            obj_class,
+            key=key_id_map.get_object_key(event.object_id),
+            class_id=event.tool_class_id,
+        )
+        vid_figure = sly.VideoFigure(
+            video_object,
+            figure.bbox,
+            event_frame_idx,
+            # key=key_id_map.get_object_key(event.object_id),
+            # class_id=event.tool_class_id,
+            track_id=event.object_id,
+        )
+        processed_vidfigures.append(vid_figure)
+    except AttributeError:
+        for frame in ann_json["frames"]:
+            if frame["index"] == event.frame:
+                for figure in frame["figures"]:
+                    object_json = next(
+                        (item for item in ann_json["objects"] if item["id"] == figure["objectId"]),
+                        None,
+                    )
+                    vid_object = sly.VideoObject.from_json(object_json, project_meta).clone(
+                        key=key_id_map.get_object_key(figure["objectId"]),
+                        class_id=object_json["classId"],
+                    )
+                    bbox = sly.Rectangle.from_json(figure["geometry"])
+                    processed_vidfigures.append(
+                        sly.VideoFigure(vid_object, bbox, event.frame, track_id=figure["objectId"])
+                    )
+                break
 
-    new_figures = []
-    drow, dcol = 0, 0
-    for idx, frame_index in enumerate(frame_indexes):
+    for video_figure in processed_vidfigures:
 
-        if idx == event_frame_idx:
-            drow, dcol = 0, 0
-            continue
+        # video_object = sly.VideoObject(
+        #     obj_class,
+        #     key=key_id_map.get_object_key(event.object_id),
+        #     class_id=event.tool_class_id,
+        # )
 
-        direction = -1 if idx < event_frame_idx else 1
-        frame_id = f"{event.video_id}_{frame_index}"
+        new_figures = []
+        drow, dcol = 0, 0
+        for idx, frame_index in enumerate(frame_indexes):
 
-        _diff = get_increment_by_diff(
-            api, event, frame_id, direction, reverse_direction=True, modality="video"
-        )
-        # if _diff is None:
-        _drow, _dcol = _diff
+            if idx == event_frame_idx:
+                drow, dcol = 0, 0
+                continue
 
-        drow += _drow * direction
-        dcol += _dcol * direction
+            direction = -1 if idx < event_frame_idx else 1
+            frame_id = f"{event.video_id}_{frame_index}"
 
-        new_figure = interpolate_video_figure(
-            video_figure, video_info, video_object, frame_index, drow, dcol
-        )
-        if new_figure is None:
-            continue
-        if (direction < 0 and new_figure.geometry.to_bbox().bottom < 0) or (
-            direction > 0 and new_figure.geometry.to_bbox().top > video_info.frame_height
-        ):
-            continue
+            _drow, _dcol = get_increment_by_diff(
+                api, event, frame_id, direction, reverse_direction=True, modality="video"
+            )
 
-        new_figures.append(new_figure)
+            drow += _drow * direction
+            dcol += _dcol * direction
 
-    api.video.figure.append_bulk(event.video_id, new_figures, key_id_map)
+            new_figure = interpolate_video_figure(
+                video_figure, video_info, video_figure.video_object, frame_index, drow, dcol
+            )
+            if new_figure is None:
+                continue
+            if (direction < 0 and new_figure.geometry.to_bbox().bottom < 0) or (
+                direction > 0 and new_figure.geometry.to_bbox().top > video_info.frame_height
+            ):
+                continue
+
+            new_figures.append(new_figure)
+
+        api.video.figure.append_bulk(event.video_id, new_figures, key_id_map)
 
-    if g.event_anns[event.user_id][frame_id] == "done":
-        g.event_anns[event.user_id][frame_id] = {}
-    g.event_anns[event.user_id][frame_id][video_figure.track_id] = video_figure
+        # if g.event_anns[event.user_id][frame_id] == "done":
+        #     g.event_anns[event.user_id][frame_id] = {}
+        # g.event_anns[event.user_id][frame_id][video_figure.track_id] = video_figure
 
 
 @app.event(sly.Event.FigureCreated)
@@ -669,6 +694,10 @@ def manual_track():
         return
     status_image_not_initialized.hide()
 
+    if getattr(event, "frame", None) is not None:
+        process_video(api, event)
+        return
+
     project_meta = g.get_project_meta(api, event.project_id)
 
     track_tag_meta = project_meta.get_tag_meta(g.TRACK_TAG_NAME)