Investigation of new techniques and development of solutions for video and image recognition under severe processing conditions with Hidden Markov models, Classical Neural Networks, Convolutional Neural Networks and Deep Learning. Solutions are investigated and proposed for different steps of image processing and computer visions systems involving image treatment, detection of ROIs, recognition, validation and interpretation. Other techniques such as High Curvature Points, Tangent Angle Signature, Wavelet Transform and Dynamic Time Warping are also considered. Different applications such as described below are object of investigation:
- Object recognition in real scenes subject to severe occlusion and noisy
- Scene analysis and interpretation in images and videos
- Computer aided diagnosis
- Computer aided patient monitoring
- Driver behavior analysis based on image and biometrics
- Integrated analysis of tect ang image for user behavior extraction and recognition