Mask r-cnn报错ValueError: Dimension 0 in both shapes must be equal, but are 7 and 64.

在Mask r-cnn预测阶段报错

报错

ValueError: Dimension 0 in both shapes must be equal, but are 7 and 64. Shapes are [7,7,4,64] and [64,3,7,7]. for 'Assign' (op: 'Assign') with input shapes: [7,7,4,64], [64,3,7,7].

图片输入格式为RGB，通道数无误，num_class确认无误，考虑图片读取问题

解决办法

参考：Annonal的博客

将

 Image= skimage.io.imread(file_name)

改为

 image = cv2.imread(file_name)

skimage.io.imread与cv2.imread保存后都是numpy格式，但cv2的储存格式是BGR，而skimage的储存格式为RGB。