Created by: zouxiaodong

训练时输入shape为3，900，900 python eval.py
--data_dir=/root/paddle/lemon/data/augmentation/
--model=ResNet50
--batch_size=8
--class_dim=3
--image_shape=3,900,900
--crop_size=900
--use_gpu=false
--pretrained_model=/root/paddle/lemon/new_train_model/ResNet50/119/

==错误信息如下： ------------- Configuration Arguments ------------- batch_size : 8 class_dim : 3 crop_size : 900 data_dir : /root/paddle/lemon/data/augmentation/ image_mean : [0.485, 0.456, 0.406] image_shape : 3,900,900 image_std : [0.229, 0.224, 0.225] interpolation : None model : ResNet50 padding_type : SAME pretrained_model : /root/paddle/lemon/new_train_model/ResNet50/119/ reader_buf_size : 2048 reader_thread : 8 resize_short_size : 256 use_gpu : 0 use_se : True

Traceback (most recent call last): File "eval.py", line 154, in main() File "eval.py", line 150, in main eval(args) File "eval.py", line 119, in eval feed=feeder.feed(data)) File "/root/miniconda3/envs/lemon/lib/python3.7/site-packages/paddle/fluid/data_feeder.py", line 296, in feed ret_dict[each_name] = each_converter.done() File "/root/miniconda3/envs/lemon/lib/python3.7/site-packages/paddle/fluid/data_feeder.py", line 105, in done arr = numpy.array(self.data, dtype=self.dtype) ValueError: could not broadcast input array from shape (3,256,297) into shape (3) terminate called without an active exception W1202 11:51:39.412518 13898 init.cc:205] *** Aborted at 1575258699 (unix time) try "date -d @1575258699" if you are using GNU date *** W1202 11:51:39.416721 13898 init.cc:205] PC: @ 0x0 (unknown) FATAL: exception not rethrown