OpenCV Python YOLOv3目标识别代码报IndexError错误求助

我正在用Python结合OpenCV做目标识别开发，基于YOLOv3模型写了如下代码：

import cv2 as cv
import numpy as np

img = cv.imread("photo1.jpg")
print(img)

img_width = img.shape[1]
img_height = img.shape[0]

img_blob = cv.dnn.blobFromImage(img,1/255,(416,416),swapRB = True)
lables = ["person", "bicycle", "car", "motorcycle", "airplane", "bus", "train", "truck", "boat",
          "trafficlight", "firehydrant", "stopsign", "parkingmeter", "bench", "bird", "cat",
          "dog", "horse", "sheep", "cow", "elephant", "bear", "zebra", "giraffe", "backpack",
          "umbrella", "handbag", "tie", "suitcase", "frisbee", "skis", "snowboard", "sportsball",
          "kite", "baseballbat", "baseballglove", "skateboard", "surfboard", "tennisracket",
          "bottle", "wineglass", "cup", "fork", "knife", "spoon", "bowl", "banana", "apple",
          "sandwich", "orange", "broccoli", "carrot", "hotdog", "pizza", "donut", "cake", "chair",
          "sofa", "pottedplant", "bed", "diningtable", "toilet", "tvmonitor", "laptop", "mouse",
          "remote", "keyboard", "cellphone", "microwave", "oven", "toaster", "sink", "refrigerator",
          "book", "clock", "vase", "scissors", "teddybear", "hairdrier", "toothbrush"]

colors = ["255,255,0","0,255,0","255,0,255","0,255,255","0,0,255"]
colors = [np.array(color.split(",")).astype("int") for color in colors]
colors = np.array(colors)
colors = np.tile(colors,(20,1))

model = cv.dnn.readNetFromDarknet ("Model/yolov3.cfg","Model/yolov3.weights")
layers = model.getLayerNames()
output_layer = [layers[layer[0]-1] for layer in model.getUnconnectedOutLayers()]

model.setInput(img_blob)

detection_layers = model.forward(output_layer)

ids_list = []
boxes_list = []
confidences_list = []

for detection_layer in detection_layers:
    for object_detection in detection_layer:
        scores = object_detection[5:]
        predicted_id = np.argmax(scores)
        confidence = scores[predicted_id]

        if confidence > 0.80:
            label = lables[predicted_id]
            bounding_box = object_detection[0:4] * np.array(img_width,img_height,img_width,img_height)
            (box_center_x, box_center_y,box_width,box_height) = bounding_box.astype("int")

            start_x = int(box_center_x - (box_width/2))
            start_y = int(box_center_y - (box_height/2))

            ids_list.append(predicted_id)
            confidences_list.append(float(confidence))
            boxes_list.append([start_x,start_y,int(box_width),int(box_height)])

max_ids = cv.dnn.NMSBoxes(boxes_list,confidences_list,0.5,0.4)

for max_id in max_ids:
    max_class_id = max_id[0]
    box = boxes_list[max_class_id]

    start_x = box[0]
    start_y = box[1]
    box_width = box[2]
    box_height = box[3]

    predicted_id = ids_list[max_class_id]
    label = lables[predicted_id]
    confidence = confidences_list[max_class_id]

    end_x = start_x + box_width
    end_y = start_y + box_height

    box_color = colors[predicted_id]
    box_color = [int(each)for each in box_color]

    cv.rectangle(img,(start_x,start_y),(end_x,end_y),box_color,2)
    cv.putText(img,label,(start_x,start_y-20),cv.FONT_HERSHEY_SIMPLEX,0.5,box_color,1)

cv.imshow("Detection Screen", img)

运行代码时出现如下错误：

Traceback (most recent call last):
  File "C:\Users\bahac\OneDrive\Desktop\OpenCV\main.py", line 29, in <module>
    output_layer = [layers[layer[0]-1] for layer in model.getUnconnectedOutLayers()]
                   ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "C:\Users\bahac\OneDrive\Desktop\OpenCV\main.py", line 29, in <listcomp>
    output_layer = [layers[layer[0]-1] for layer in model.getUnconnectedOutLayers()]
                           ~~~~~^^^
IndexError: invalid index to scalar variable.

解决方案 #

1. 核心错误原因 #

报错源于OpenCV版本差异导致的返回格式变化：

• 旧版OpenCV中，model.getUnconnectedOutLayers()返回嵌套数组（如[[200], [227], [254]]）
• 新版OpenCV中，该方法返回一维数组（如[200, 227, 254]）
  原代码中layer[0]的写法只适配旧版，新版中layer是标量值，无法使用索引，因此触发IndexError。

2. 代码修复 #

将获取输出层的代码替换为兼容新旧版本的写法：

layers = model.getLayerNames()
unconnected_layers = model.getUnconnectedOutLayers()
# 统一处理为一维数组，再遍历取值
output_layer = [layers[i - 1] for i in unconnected_layers.flatten()]

3. 额外潜在问题修复 #

代码中计算bounding_box的行存在语法错误，缺少数组方括号，会触发TypeError，需修改：

# 原代码：
bounding_box = object_detection[0:4] * np.array(img_width,img_height,img_width,img_height)
# 修改为：
bounding_box = object_detection[0:4] * np.array([img_width, img_height, img_width, img_height])

4. 验证 #

修改后重新运行代码，即可正常加载模型、获取输出层并执行目标检测。

内容的提问来源于stack exchange，提问作者İbrahim Ethem BAHACAN