Image processing | Pattern recognition

The project aims to pedestrian detection. The database for training, testing and evaluating the results will be INRIA. It can be downloaded from here : [login to view URL] The files are in .vbb format but i can give you the extracted images, the matlab code for reading and evaluating the database and the HOG algorithm that extractes the features. Through the annotations the pedestrians in every train image must be croped and extract HOG features. HOG method is explained here : [login to view URL] Later on based on these features you must train a deep belief network (supervised learning) to extract the coefficients of the network. After, you must evaluate the results of testing images (miss rate). Additional papers (help) : [login to view URL] [login to view URL]