add object crop

app.py

@@ -227,6 +227,12 @@ def yolo_det_img(img, device, model_name, infer_size, conf, iou, max_num, model_
     img_size = img.size  # frame size
 
     results = model(img, size=infer_size)  # detection
+    
+    # ----------------目标裁剪----------------
+    crops = results.crop(save=False)
+    img_crops = []
+    for i in range(len(crops)):
+        img_crops.append(crops[i]["im"][..., ::-1])
 
     # Data Frame
     dataframe = results.pandas().xyxy[0].round(2)
@@ -320,7 +326,7 @@ def yolo_det_img(img, device, model_name, infer_size, conf, iou, max_num, model_
     for k, v in clsDet_dict.items():
         clsRatio_dict[k] = v / clsDet_dict_sum
 
-    return det_img, objSize_dict, clsRatio_dict, det_json, report, dataframe
+    return det_img, objSize_dict, clsRatio_dict, det_json, report, dataframe, img_crops
 
 
 # YOLOv5 video detection function
@@ -517,12 +523,13 @@ def main(args):
                               label="List of detection information")
     outputs_objSize = gr.Label(label="Object size ratio statistics")
     outputs_clsSize = gr.Label(label="Category detection proportion statistics")
+    outputs_crops = gr.Gallery(label="Object crop")
 
     # -------------------output component-------------------
     outputs_video = gr.Video(format='mp4', label="Detection video")
 
     # output parameters
-    outputs_img_list = [outputs_img, outputs_objSize, outputs_clsSize, outputs_json, outputs_pdf, outputs_df]
+    outputs_img_list = [outputs_img, outputs_objSize, outputs_clsSize, outputs_json, outputs_pdf, outputs_df, outputs_crops]
     outputs_video_list = [outputs_video]
 
     # title