deeplabv3+如何自定义类别?
Created by: 408550969
我换了一个数据集,将train.py的num_classes改成10并把models.py里的label_number改成10。但是运行就报错: File "./train.py", line 148, in 'label': labels}) File "/home/cll/.local/lib/python2.7/site-packages/paddle/fluid/parallel_executor.py", line 277, in run self.executor.run(fetch_list, fetch_var_name) paddle.fluid.core.EnforceNotMet: Enforce failed. Expected output_shape[unk_dim_idx] * capacity == -in_size, but received output_shape[unk_dim_idx] * capacity:-11235850 != -in_size:-11235859. Invalid shape is given. at [/paddle/paddle/fluid/operators/reshape_op.cc:98] PaddlePaddle Call Stacks: 而且我发现只有调整为默认的19它才不报错,请问应该如何修改? 还有train.py里的loss label_nignore = (label < num_classes).astype('float32') 这个是什么意思?