caffemodel转PaddlePaddle 失败 AttributeError: input

[78666621]

模型来源： CAFFE 模型说明： RCF边缘检测，其相应GitHub：https://github.com/yun-liu/RCF 模型文件：链接：https://pan.baidu.com/s/1vscZPDIWE3tia9V-1Fo7OQ 提取码：c2ya 环境：python3.8 转换过程出错提示如下： paddle.version = 2.1.1 Now translating model from caffe to paddle. Total nodes: 70 Nodes converting ... Converting node 15 ... Traceback (most recent call last): File "G:\Anaconda3\envs\eiseg\Scripts\x2paddle-script.py", line 33, in sys.exit(load_entry_point('x2paddle==1.3.1', 'console_scripts', 'x2paddle')()) File "G:\Anaconda3\envs\eiseg\lib\site-packages\x2paddle\convert.py", line 255, in main caffe2paddle(args.prototxt, args.weight, args.save_dir, File "G:\Anaconda3\envs\eiseg\lib\site-packages\x2paddle\convert.py", line 137, in caffe2paddle mapper = CaffeOpMapper(model) File "G:\Anaconda3\envs\eiseg\lib\site-packages\x2paddle\op_mapper\caffe2paddle\caffe_op_mapper.py", line 147, in init self.directly_map(node) File "G:\Anaconda3\envs\eiseg\lib\site-packages\x2paddle\op_mapper\caffe2paddle\caffe_op_mapper.py", line 171, in directly_map inputs = node.layer.input AttributeError: input

[yeliang2258]

https://github.com/PaddlePaddle/X2Paddle/pull/663 你好，由于X2Paddle先前Eltwise和Crop, 以及Directly map 3个OP类型转换过程中字段没考虑完全导致转换失败，目前在PR中已修复，请拉这个PR进行测试。