The Status and Challenges of Image Data Augmentation Algorithms

Abstract Image data-augmentation algorithms effectively trump the problem of insufficient training samples for deep learning in some application fields, and it is typically scholars to choose them various computer vision tasks. But as develop rapidly, early proposed classification that are sorted into classical ways generating methods no more suitable, because such misses other meaningful strategies. Besides, frustrating someone decide which exact method undertake, though there too many optional choose. Towards goal making suggestions, paper categorizes image three kinds from perspective algorithm strategy, they matrix transformation algorithm, feature expansion model generation based on neural network algorithm. The analyzes typical principle, performance, scenarios, research status future challenges, forecasts development trend data augmentation algorithms. can provide academic reference fields medicine military.