Andrew:
● Data augmentation is quite an advanced technique, and might be too much for absolute beginners. A lot can go wrong and its very context specific.
● I think more should be made of data quality. Im not sure it would be necessary to go into depth around normalisation, standardisation ,image pipelines, etc. but data quality and consistency is absolutely key, and for most projects a bigger issue than the ML techniques themselves.
Andrew: ● Data augmentation is quite an advanced technique, and might be too much for absolute beginners. A lot can go wrong and its very context specific. ● I think more should be made of data quality. Im not sure it would be necessary to go into depth around normalisation, standardisation ,image pipelines, etc. but data quality and consistency is absolutely key, and for most projects a bigger issue than the ML techniques themselves.