YOLOv8obb获取边框坐标时报错AttributeError: 'NoneType' object has no attribute 'xywhr'

Question

使用YOLOv8obb模型获取边框坐标时报错obbs = r.boxes.xywhr.cpu().tolist()
AttributeError: 'NoneType' object has no attribute 'xywhr'

import time
import cv2
from ultralytics import YOLO
from ultralytics.utils.plotting import save_one_box as crop
import imutils
import numpy as np
import os
import shutil

model = YOLO('best.pt')

video_path ="D:/WIN_20230907_21_08_23_Pro.mp4"
cap = cv2.VideoCapture(video_path)
confi=0.2
while cap.isOpened():
    success, frame = cap.read()

    confi = round(confi, 2)
    if confi <= 0.2:
        confi = 0.2
    elif confi >= 0.8:
        confi = 0.8

    if success:
        results = model(frame,  conf=confi, save=False, show=False)
        annotated_frame = results[0].plot()
        cv2.imshow("YOLOv8 Tracking", annotated_frame)
        r=results[0]

        if results:
            obbs = r.boxes.xywhr.cpu().tolist()
            for i, obb in enumerate(obbs):
                xc, yc, w, h, angle = obb  # Unpack the OBB

        else:
            bkg = np.ones((256, 1280, 3), dtype=np.uint8)
            bkg *= 200
            annotated_frame = np.vstack((annotated_frame, bkg))

        if cv2.waitKey(1) & 0xFF == ord("q"):
            break
    else:
        # Break the loop if the end of the video is reached
        break

    # Release the video capture object and close the display window
cap.release()
cv2.destroyAllWindows()

修改后不管有没有检测到物体，都会print：no objects detected
最前面有这个报错不知道有没有关系：

UserWarning: Plan failed with a cudnnException: CUDNN_BACKEND_EXECUTION_PLAN_DESCRIPTOR: cudnnFinalize Descriptor Failed cudnn_status: CUDNN_STATUS_NOT_SUPPORTED (Triggered internally at C:\actions-runner\_work\pytorch\pytorch\builder\windows\pytorch\aten\src\ATen\native\cudnn\Conv_v8.cpp:919.)
  return F.conv2d(input, weight, bias, self.stride,

Answer 1

参考免费微信小程序：皆我百晓生 作答，经由专家-皆知整理回复，如问题解决，望采纳；如仍有疑问，请评论回复。

遇到这个错误AttributeError: 'NoneType' object has no attribute 'xywhr'通常意味着results[0]没有成功检测到任何对象，因此r.boxes是None，尝试访问.xywhr属性自然会引发错误。为了解决这个问题，你需要在尝试访问边界框属性之前，先检查是否有检测结果。下面是对你的代码进行修改后的版本，加入了对无检测结果情况的处理：

import time
import cv2
from ultralytics import YOLO
from ultralytics.utils.plotting import save_one_box as crop
import imutils
import numpy as np
import os
import shutil

# Load the model
model = YOLO('best.pt')

video_path = "D:/WIN_20230907_21_08_23_Pro.mp4"
cap = cv2.VideoCapture(video_path)
confi = 0.2

while cap.isOpened():
    success, frame = cap.read()

    # Ensure confidence is within bounds
    confi = max(min(confi, 0.8), 0.2)

    if success:
        results = model(frame, conf=confi, save=False, show=False)
        
        # Annotate the frame only if there are detections
        if results:
            annotated_frame = results[0].plot()
            cv2.imshow("YOLOv8 Tracking", annotated_frame)
            
            # Access boxes after confirming detections exist
            r = results[0]
            if r.boxes:
                obbs = r.boxes.xywhr.cpu().tolist()
                for i, obb in enumerate(obbs):
                    xc, yc, w, h, angle = obb  # Unpack the OBB
                    # Process each bounding box here...
            else:
                print("No detections found for this frame.")
        else:
            # If no detections, create a blank background and stack it
            bkg = np.ones((256, 1280, 3), dtype=np.uint8) * 200
            annotated_frame = np.vstack((annotated_frame, bkg))
            cv2.imshow("YOLOv8 Tracking", annotated_frame)

        # Exit condition
        if cv2.waitKey(1) & 0xFF == ord("q"):
            break
    else:
        break

# Cleanup
cap.release()
cv2.destroyAllWindows()

这段代码中，我做了以下修改：

1. 在调整confi值时使用了max和min函数来确保其始终在0.2到0.8之间。
2. 在尝试访问r.boxes.xywhr之前，首先检查了results是否非空，然后进一步检查了r.boxes是否存在，以避免尝试访问NoneType对象的属性。
3. 对于没有检测结果的帧，添加了一条打印语句来提示，并且仍然显示了画面（即使没有实际检测结果，也通过叠加空白背景来保持界面更新）。

这样修改后，你的程序应该能够在没有检测到任何对象时避免抛出异常，并且能继续正常运行。