I was fine tuning LFFD on my custom dataset for pedestrians where my input shape was (640,480). After logging parameters training script gave the following error:
2020-05-26 11:52:08,650[INFO]: -----------------------------------------------------------------------------------
infer_shape error. Arguments:
Traceback (most recent call last):
File "config_farm/configuration_30_320_20L_4scales_v1.py", line 312, in <module>
run()
File "config_farm/configuration_30_320_20L_4scales_v1.py", line 308, in run
start_index=param_start_index)
File "/home/nagarro/work/curved_text_detection/send/A-Light-and-Fast-Face-Detector-for-Edge-Devices/ChasingTrainFramework_GeneralOneClassDetection/train_GOCD.py", line 60, in start_train
solver.fit()
File "/home/nagarro/work/curved_text_detection/send/A-Light-and-Fast-Face-Detector-for-Edge-Devices/ChasingTrainFramework_GeneralOneClassDetection/solver_GOCD.py", line 90, in fit
self.__init_module()
File "/home/nagarro/work/curved_text_detection/send/A-Light-and-Fast-Face-Detector-for-Edge-Devices/ChasingTrainFramework_GeneralOneClassDetection/solver_GOCD.py", line 63, in __init_module
arg_shapes, _, __ = self.net_symbol.infer_shape()
File "/usr/local/lib/python3.6/dist-packages/mxnet/symbol/symbol.py", line 1103, in infer_shape
res = self._infer_shape_impl(False, *args, **kwargs)
File "/usr/local/lib/python3.6/dist-packages/mxnet/symbol/symbol.py", line 1267, in _infer_shape_impl
ctypes.byref(complete)))
File "/usr/local/lib/python3.6/dist-packages/mxnet/base.py", line 255, in check_call
raise MXNetError(py_str(_LIB.MXGetLastError()))
mxnet.base.MXNetError: Error in operator conv11_loss_score: Shape inconsistent, Provided = [16,2,59,59], inferred shape=[16,2,59,79]
Also, did you train on (640,480) or (480,480) because your training script had (480,480) by default but pickle files were of dimensions (640,480).
I was fine tuning LFFD on my custom dataset for pedestrians where my input shape was (640,480). After logging parameters training script gave the following error:
Also, did you train on (640,480) or (480,480) because your training script had (480,480) by default but pickle files were of dimensions (640,480).