I try testing disparities_pp.npy with --do_stereo flag.
When I tested disparities_pp.npy without --do_stereo flag, the results are good.
After training with --do_stereo flag, the evaluation results of disparities.npy looked good.
However, disparities_nn.npy results became worse : the error rate went worst boundary.
Do I have to do something when to evaluate disparities_pp.npy file with do_stereo flag?
I try testing disparities_pp.npy with --do_stereo flag. When I tested disparities_pp.npy without --do_stereo flag, the results are good. After training with --do_stereo flag, the evaluation results of disparities.npy looked good. However, disparities_nn.npy results became worse : the error rate went worst boundary.
Do I have to do something when to evaluate disparities_pp.npy file with do_stereo flag?