DiscrimNet: Semi-Supervised Action Recognition from Videos using Generative Adversarial Networks

We propose an action recognition framework using Generative Adversarial Networks. Our model involves training a deep convolutional generative adversarial network (DCGAN) using a large video activity dataset without label information. Then we use the trained discriminator from the GAN model as an unsupervised pre-training step and fine-tune the trained discriminator model on a labeled dataset to recognize human activities. We determine good network architectural and hyperparameter settings for using the discriminator from DCGAN as a trained model to learn useful representations for action recognition. Our semi-supervised framework using only appearance information achieves superior or comparable performance to the current state-of-the-art semi-supervised action recognition methods on two challenging video activity datasets: UCF101 and HMDB51.