Our team is working on two projects which aim at understanding the human attention for visual object search.
Attention Model for Object Detection
- In this we are using RL-SOM trained with temporal difference error to develop a attentional search model. The model currently trained on on synthetic data, with the broder goal to implement it on road video dataset to search traffic signs. This project is done in collaboration with Continental Automotive.
Developing an Human Gaze Tracker
- We are trying to record the human gaze to understand how eye movements help in searching object. For this purpose, we are developing an human gaze tracker using a webcam. The data collected will be used to train a model which can then search objects like humans do.