在rknn-toolkit:1.4.0的docker里caffemodel转rknn

[复制链接] · 发表于 2020-12-23 17:06:16

W Rename slice_[dump, mask] to rknn_slice_[dump, mask]

W Unknown caffe layer Silence_[dump_blob], op is silence, please REMOVE it from prototxt if possible.

W Rename Silence_[dump_blob] to rknn_Silence_[dump_blob]

W Rename Slice_['category_pt', 'instance_pt', 'confidence_pt', 'classify_pt', 'heading_pt', 'height_pt'] to rknn_Slice_['category_pt', 'instance_pt', 'confidence_pt', 'classify_pt', 'heading_pt', 'height_pt']

W Rename attach_rknn_Slice_['category_pt', 'instance_pt', 'confidence_pt', 'classify_pt', 'heading_pt', 'height_pt']_56 to rknn_attach_rknn_Slice_['category_pt', 'instance_pt', 'confidence_pt', 'classify_pt', 'heading_pt', 'height_pt']_56

W Rename attach_rknn_Slice_['category_pt', 'instance_pt', 'confidence_pt', 'classify_pt', 'heading_pt', 'height_pt']_56 to rknn_attach_rknn_Slice_['category_pt', 'instance_pt', 'confidence_pt', 'classify_pt', 'heading_pt', 'height_pt']_56

W Rename attach_rknn_Slice_['category_pt', 'instance_pt', 'confidence_pt', 'classify_pt', 'heading_pt', 'height_pt']_56 to rknn_attach_rknn_Slice_['category_pt', 'instance_pt', 'confidence_pt', 'classify_pt', 'heading_pt', 'height_pt']_56

E Catch exception when loading caffe model: /home/deploy.prototxt!

E Traceback (most recent call last):

E   File "rknn/api/rknn_base.py", line 506, in rknn.api.rknn_base.RKNNBase.load_caffe

E   File "rknn/base/RKNNlib/switcher/mode_switcher.py", line 347, in rknn.base.RKNNlib.switcher.mode_switcher.C2T_Switcher.switch

E   File "rknn/base/RKNNlib/switcher/mode_switcher.py", line 315, in rknn.base.RKNNlib.switcher.mode_switcher.ModeSwitcher.build_mark

E   File "rknn/base/RKNNlib/switcher/mode_switcher.py", line 191, in rknn.base.RKNNlib.switcher.mode_switcher.ModeSwitcher.optimize_mark

E   File "rknn/base/RKNNlib/switcher/mark_optimize.py", line 449, in rknn.base.RKNNlib.switcher.mark_optimize.MarkOptimize.apply

E   File "rknn/base/RKNNlib/switcher/mark_optimize.py", line 445, in rknn.base.RKNNlib.switcher.mark_optimize.MarkOptimize.forward_mark

E   File "rknn/base/RKNNlib/switcher/mark_optimize.py", line 28, in rknn.base.RKNNlib.switcher.mark_optimize.fwd_upstream

E AttributeError: 'NoneType' object has no attribute 'reset_shape_format'
复制代码

这是因为有特殊层吗？

只看该作者 · 发表于 2020-12-28 14:23:02

删掉那个未知类型的层就可以了