【发布时间】：2021-06-30 14:47:21
【问题描述】：
我已经在 google colab 上的自定义数据集上训练了 yolo-tiny-v4，并且检测效果很好。然后我尝试在opencv的dnn模块的帮助下将yolo-tiny-v4加载到其他colabproject中，没有出现错误，但是检测失败（没有检测到物体，检测的输出是Nan的向量） .
[array([[nan, nan, nan, nan, nan, 0.],
[nan, nan, nan, nan, nan, 0.],
[nan, nan, nan, nan, nan, 0.],
...,
[nan, nan, nan, nan, nan, 0.],
[nan, nan, nan, nan, nan, 0.],
[nan, nan, nan, nan, nan, 0.]], dtype=float32),
array([[nan, nan, nan, nan, nan, 0.],
[nan, nan, nan, nan, nan, 0.],
[nan, nan, nan, nan, nan, 0.],
...,
[nan, nan, nan, nan, nan, 0.],
[nan, nan, nan, nan, nan, 0.],
[nan, nan, nan, nan, nan, 0.]], dtype=float32)]
我在 colab 上使用 OpenCV 版本 4.5.1 和 Python 3.7。有什么想法吗？
这是代码
#Load YOLO
net = cv2.dnn.readNetFromDarknet("/content/custom-yolov4-tiny-detector.cfg","/content/custom-yolov4-
tiny-detector_last.weights")
classes = []
with open("obj.names","r") as f:
classes = [line.strip() for line in f.readlines()]
net.getLayerNames()
layer_names = net.getLayerNames()
outputlayers = [layer_names[i[0] - 1] for i in net.getUnconnectedOutLayers()]
#loading image
img = cv2.imread("/content/1.png")
#img = cv2.resize(img,None,fx=0.4,fy=0.3)
height,width,channels = img.shape
cv2_imshow(img)
blob = cv2.dnn.blobFromImage(img,0.00392,(416,416),(0,0,0),True,crop=False)
net.setInput(blob)
outs = net.forward(outputlayers)
outs
【问题讨论】：
-
相同的代码是否适用于预训练的微型 yolo v3 模型？您是否尝试过预训练的微型 yolo v4？
-
我会试试这个并提供反馈
-
@Micka 嗨，我犯了一个有趣的错误，因为初学者忘记在可乐中安装暗网。现在一切正常，谢谢
标签： opencv computer-vision conv-neural-network yolo transfer-learning