atualizando

app.py

@@ -1,7 +1,4 @@
-import ultralytics
-import onemetric
 import supervision
-import typing
 import tqdm
 import os
 from ultralytics import YOLO
@@ -14,7 +11,6 @@ from supervision import Color
 from supervision import VideoInfo
 from supervision import get_video_frames_generator
 from supervision import VideoSink
-import torch
 os.system("pip install git+https://github.com/ifzhang/ByteTrack")
 from typing import List
 import numpy as np
@@ -90,7 +86,13 @@ def ObjectDetection(video_path):
     video_info = VideoInfo.from_video_path(video_path)
     generator = get_video_frames_generator(video_path)
     box_annotator = BoxAnnotator(thickness=5, text_thickness=5, text_scale=1)
-
+    #polygon
+    polygon = np.array([[200,300], [200,1420], [880, 1420], [880, 300]])
+    #zone
+    zone = supervision.PolygonZone(polygon=polygon, frame_resolution_wh=video_info.resolution_wh)
+    #zone annotator
+    zone_annotator = supervision.PolygonZoneAnnotator(zone=zone, color=Color.white(), thickness=4)
+    # open target video file
     with VideoSink(TARGET_VIDEO_PATH, video_info) as sink:
         # loop over video frames
         for frame in tqdm(generator, total=video_info.total_frames):
@@ -119,10 +121,14 @@ def ObjectDetection(video_path):
                 in detections
             ]
             t = np.unique(detections.class_id, return_counts =True)
-            for x in zip(t[0], t[1]):
-                frame = draw_text(background_color=Color.white(), scene=frame, text=' '.join((str(classes[x[0]]), ':', str(x[1]))), text_anchor=Point(x=50, y=300 + (50 * x[0])), text_scale = 2, text_thickness = 4, )
             # annotate and display frame
-            frame = box_annotator.annotate(scene=frame, detections=detections, labels=labels)
+            mask = zone.trigger(detections=detections)
+            detections_filtered = detections[mask]
+            t = np.unique(detections_filtered.class_id, return_counts =True)
+            for x in zip(t[0], t[1]):
+                frame = draw_text(background_color=Color.white(), scene=frame, text=' '.join((str(classes[x[0]]), ':', str(x[1]))), text_anchor=Point(x=500, y=1550 + (50 * x[0])), text_scale = 2, text_thickness = 4)
+            frame = box_annotator.annotate(scene=frame, detections=detections_filtered, labels=labels)
+            frame = zone_annotator.annotate(scene=frame)
             sink.write_frame(frame)
 
     return TARGET_VIDEO_PATH