Created by: eddieheyutong

如题，我在通过export_model.py文件导出了下载的Paddle 官方预训练模型，deeplabv3p_xception65_bn_cityscapes，然后用之前部署的C++预测库进行预测的时候，报出以下错误 /root/projects/PaddleSeg/inference/build# ./demo --conf=/root/projects/PaddleSeg/inference/conf/deploy.yaml --input_dir=/root/projects/PaddleSeg/inference/images/cityscapes/ WARNING: Logging before InitGoogleLogging() is written to STDERR I1015 16:15:59.258088 19235 analysis_predictor.cc:88] Profiler is deactivated, and no profiling report will be generated. I1015 16:15:59.278381 19235 op_compatible_info.cc:201] The default operator required version is missing. Please update the model version. I1015 16:15:59.278416 19235 analysis_predictor.cc:841] MODEL VERSION: 0.0.0 I1015 16:15:59.278431 19235 analysis_predictor.cc:843] PREDICTOR VERSION: 0.0.0 W1015 16:15:59.278589 19235 analysis_predictor.cc:855] - Version incompatible (1) batch_norm W1015 16:15:59.278600 19235 analysis_predictor.cc:855] - Version incompatible (1) bilinear_interp W1015 16:15:59.278607 19235 analysis_predictor.cc:855] - Version incompatible (1) concat W1015 16:15:59.278614 19235 analysis_predictor.cc:855] - Version incompatible (1) conv2d W1015 16:15:59.278622 19235 analysis_predictor.cc:855] - Version incompatible (1) depthwise_conv2d W1015 16:15:59.278630 19235 analysis_predictor.cc:855] - Version incompatible (1) dropout W1015 16:15:59.278637 19235 analysis_predictor.cc:855] - Version incompatible (1) elementwise_add W1015 16:15:59.278646 19235 analysis_predictor.cc:855] - Version incompatible (1) feed W1015 16:15:59.278652 19235 analysis_predictor.cc:855] - Version incompatible (1) fetch W1015 16:15:59.278659 19235 analysis_predictor.cc:855] - Version incompatible (1) reduce_mean W1015 16:15:59.278667 19235 analysis_predictor.cc:855] - Version incompatible (1) relu W1015 16:15:59.278674 19235 analysis_predictor.cc:855] - Version incompatible (1) scale W1015 16:15:59.278682 19235 analysis_predictor.cc:855] - Version incompatible (1) softmax W1015 16:15:59.278688 19235 analysis_predictor.cc:855] - Version incompatible (1) transpose2 W1015 16:15:59.278705 19235 analysis_predictor.cc:148] WARNING: Results may be DIFF! Using same versions between model and lib. --- Running analysis [ir_graph_build_pass] --- Running analysis [ir_graph_clean_pass] --- Running analysis [ir_analysis_pass] --- Running IR pass [is_test_pass] --- Running IR pass [simplify_with_basic_ops_pass] --- Running IR pass [fc_fuse_pass] --- Running IR pass [fc_elementwise_layernorm_fuse_pass] --- Running IR pass [conv_affine_channel_fuse_pass] --- Running IR pass [conv_eltwiseadd_affine_channel_fuse_pass] --- Running IR pass [conv_bn_fuse_pass] --- detected 78 subgraphs --- Running IR pass [conv_eltwiseadd_bn_fuse_pass] --- Running IR pass [conv_elementwise_add_act_fuse_pass] --- detected 53 subgraphs --- Running IR pass [conv_elementwise_add2_act_fuse_pass] --- Running IR pass [conv_elementwise_add_fuse_pass] --- detected 26 subgraphs --- Running IR pass [transpose_flatten_concat_fuse_pass] --- Running IR pass [runtime_context_cache_pass] --- Running analysis [ir_params_sync_among_devices_pass] I1015 16:16:00.083369 19235 ir_params_sync_among_devices_pass.cc:41] Sync params from CPU to GPU --- Running analysis [adjust_cudnn_workspace_size_pass] --- Running analysis [inference_op_replace_pass] --- Running analysis [ir_graph_to_program_pass] I1015 16:16:00.203075 19235 analysis_predictor.cc:473] ======= optimize end ======= --- skip [feed], feed -> image --- skip [save_infer_model/scale_0], fetch -> fetch W1015 16:16:00.567735 19235 device_context.cc:235] Please NOTE: device: 0, CUDA Capability: 61, Driver API Version: 10.1, Runtime API Version: 10.0 W1015 16:16:00.571164 19235 device_context.cc:243] device: 0, cuDNN Version: 7.6. runtime = 14236 size of outputs[0]: (1,19,512,512,) E1015 16:16:01.581390 19235 seg_predictor.cpp:74] [FATAL] unequal: input vs output [7864320|4980736] runtime = 10389 size of outputs[0]: (1,19,512,512,) E1015 16:16:01.682529 19235 seg_predictor.cpp:74] [FATAL] unequal: input vs output [7864320|4980736] runtime = 8927 size of outputs[0]: (1,19,512,512,) E1015 16:16:01.771071 19235 seg_predictor.cpp:74] [FATAL] unequal: input vs output [7864320|4980736] runtime = 9973 size of outputs[0]: (1,19,512,512,) E1015 16:16:01.857481 19235 seg_predictor.cpp:74] [FATAL] unequal: input vs output [7864320|4980736] runtime = 6773 size of outputs[0]: (1,19,512,512,) E1015 16:16:01.934250 19235 seg_predictor.cpp:74] [FATAL] unequal: input vs output [7864320|4980736] runtime = 6660 size of outputs[0]: (1,19,512,512,) E1015 16:16:02.018646 19235 seg_predictor.cpp:74] [FATAL] unequal: input vs output [7864320|4980736]

然后图片的文件夹中并没有生成预测的图片，想请问下大佬们这个是什么原因呢？