Constrained-size Tensorflow Models for YouTube-8M Video Understanding Challenge
This work addresses the problem of efficient video classification for large-scale datasets, but it is incremental as it builds on existing methods from prior competitions.
The paper tackled the YouTube-8M video understanding challenge by building a constrained-size model that achieved 88.324% GAP on the private leaderboard, with a 48.5% compression rate using float16 precision without loss of accuracy.
This paper presents our 7th place solution to the second YouTube-8M video understanding competition which challenges participates to build a constrained-size model to classify millions of YouTube videos into thousands of classes. Our final model consists of four single models aggregated into one tensorflow graph. For each single model, we use the same network architecture as in the winning solution of the first YouTube-8M video understanding competition, namely Gated NetVLAD. We train the single models separately in tensorflow's default float32 precision, then replace weights with float16 precision and ensemble them in the evaluation and inference stages., achieving 48.5% compression rate without loss of precision. Our best model achieved 88.324% GAP on private leaderboard. The code is publicly available at https://github.com/boliu61/youtube-8m