felixBrave / ocr_chinese

图像文字检测模型(EAST/AdvancedEAST),及包含文字识别模型(CRNN+CTC),Keras/TensorFlow实现.
186 stars 44 forks source link

使用了预训练的参数但是shape是不是有点问题 #6

Closed JulianGuSM closed 2 years ago

JulianGuSM commented 4 years ago

我是报这个错误 ValueError: Dimension 1 in both shapes must be equal, but are 5994 and 5990. Shapes are [512,5994] and [512,5990]. for 'Assign_110' (op: 'Assign') with input shapes: [512,5994], [512,59 90]

snialsss commented 4 years ago

哥们解决的了没不知道怎么修改

15026529688 commented 3 years ago

同问 这个问题是什么意思

SoftCharacter commented 2 years ago

兄弟解决了吗