PROJECT TITLE :
Person Reidentification With Reference Descriptor
Person identification across nonoverlapping cameras, also known as person reidentification, aims to match individuals at completely different times and locations. Reidentifying individuals is of nice importance in crucial applications such as wide-area surveillance and visual tracking. Due to the appearance variations in cause, illumination, and occlusion in numerous camera views, person reidentification is inherently troublesome. To address these challenges, a reference-based method is proposed for person reidentification across totally different cameras. Instead of directly matching individuals by their look, the matching is conducted during a reference house where the descriptor for an individual is translated from the first color or texture descriptors to similarity measures between this person and therefore the exemplars within the reference set. A subspace is 1st learned in which the correlations of the reference knowledge from completely different cameras are maximized using regularized canonical correlation analysis (RCCA). For reidentification, the gallery knowledge and also the probe knowledge are projected onto this RCCA subspace and also the reference descriptors (RDs) of the gallery and probe are generated by computing the similarity between them and also the reference knowledge. The identity of a look is decided by comparing the RD of the probe and the RDs of the gallery. A reranking step is added to additional improve the results employing a saliency-primarily based matching theme. Experiments on publicly available datasets show that the proposed technique outperforms most of the state-of-the-art approaches.
Did you like this research project?
To get this research project Guidelines, Training and Code... Click Here