Deep neural networks for video frame prediction
keywords DEEP LEARNING, VIDEO ANALYSIS
Reference persons ENRICO MAGLI
Thesis type RESEARCH
Description Deep neural networks have proven to be tremendously powerful at modeling complex signals such as images and video sequences. The objective of this thesis is to implement and test a neural network that takes as input a set of past frames and predicts the next frame. This has applications in video analysis (e.g., event detection) and video compression.
The neural network will be based on a generative adversarial network, which is able to map video frames to a semantically meaningful latent space, and vice versa. The idea is to project past frames onto a latent space, to track the resulting features and to generate the next frame from such features.
Required skills Candidate students should have a background in ICT or mathematics. Knowledge of neural networks (including TensorFlow environment and Python programming) is not a prerequisite, although it would help in the initial stage of the activity.
Deadline 11/07/2018 PROPONI LA TUA CANDIDATURA