This code constructs action recognition models on a subset of the UCF101 dataset. The UCF101 dataset consists of a total of 13,320 videos spanning diverse human action categories. However, for this project, my focus narrows down to a subset of the dataset, comprising 2,355 training videos and 969 test videos distributed among 30 action categories. Included in the code below is:
- Binary Classifier
- Multiclass Classifier