Expand Training Data: Improving AI with Simulated Edge Cases

Most camera footage used for AI training is sunny daytime footage, because that's what cameras capture most often.

3 min readMachine Learning
From Machine Learning

I have this idea I'd like feedback on. most camera footage for training, is sunny daytime, because that's what cameras record most of the time. The edge cases models actually struggles with, like night, fog, rain or glare, are rare in the data - the idea is to augment it to have more of the rare cases for training

Read the original at Machine Learning