Sparse Modeling and Estimation of Motion Fields
1. Management and Dissemination
This task aims at controlling the project execution so that the work flow proceeds according to the schedule.
2. Sparse Estimation of Multiple Motion
First we have to devise efficient methods to estimate the model parameters. Since the number of parameters is very high (many
hundreds) efficient representations are required. This task explores the use of sparse representations of the motion fields and switching matrices.
3. Online Motion Field Estimation
Previous research (ARGUS project) dealt with offline learning of motion fields. In this project, we propose online learning to overcome both difficulties. This will be achieved in two main stages: i) Single motion field estimation;
ii) Multiple motion field estimation
4. Multicamera Systems
In this task we will tackle the extension of the motionfield paradigm to the modeling of multiple camera systems and exploit sparse methods for learning their topology and intercamera spatiotemporal relations. Learning will be performed in datasets of distributed camera networks: i) Motion fields in Multicamera Systems; ii) Learning Network Topology; iii) Cameras with overlapping fields of view; iv) Cameras with nonoverlapping
5. Activity Recognition 01012018
This task will embrace the following goals: i) Development of online inference algorithms for classifying activities in multicamera systems according to the learned models; ii) Development of methods to cope with persons' interactions; iii) Evaluate the impact of sparsity in activity recognition. We will compare different activity recognition and model learning methods, with and without sparsity constraints.
6. Nonlinear System Identification with Sparse Models
This task aims at exploiting the interplay between methods that lead to sparse models for nonlinear system identification and the
ones considered in the project for the analysis of video sequences, relying on multiple models and sparse estimation.
7. Final Report
In this task, the final report is that reports a synthesis of the main results obtained in the project is produced and a synthesis of
the work produced during the project that yielded them