Learning immediately: https://edu.csdn.net/course/play/26257/326448?utm_source=blogtoedu
labelme use open source
Supervised data enhancement:
1. Sample data enhancement: SMOT, reduce the sample is not balanced.
2.Sample Pairing
3.Mixup, linear interpolation
Generating model: the discriminator generator (to have a larger training data sets)
Autoaugmention learning has enhanced data combining operation, different tasks require different data enhancement operation.
Data enhancement can enhance the generalization ability of the model.
The internet:
1. convolution + BN + ReLU (3 layers)
2.
