|
本帖最后由 engin 于 2019-3-27 11:14 编辑
模型输入需要4通道的图像,config中增加一通道的方式报错:
rknn.config(channel_mean_value='103.94 116.78 123.68 12 58.82', reorder_channel='2 1 0 3')
不量化可以转成功,量化报错:- E Catch exception when building RKNN model!
- T Traceback (most recent call last):
- T File "rknn/api/rknn_base.py", line 470, in rknn.api.rknn_base.RKNNBase.build
- T File "rknn/api/rknn_base.py", line 889, in rknn.api.rknn_base.RKNNBase._quantize
- T File "rknn/base/rknnlib/app/tensorzone/workspace.py", line 176, in rknn.base.rknnlib.app.tensorzone.workspace.Workspace.init_image_process
- T File "rknn/base/rknnlib/app/tensorzone/workspace.py", line 185, in rknn.base.rknnlib.app.tensorzone.workspace.Workspace.init_image_process
- T File "rknn/base/rknnlib/app/tensorzone/workspace.py", line 85, in rknn.base.rknnlib.app.tensorzone.workspace.proc_tensor
- T UnboundLocalError: local variable 'scale' referenced before assignment
|
|