Re-Identification is the problem of associating identities to detections of people over a network of cameras. Occlusions, changes in illumination conditions, different camera settings, view angles and pose, are visual contingencies that contribute to make re-identification a challenging problem in videosurveillance systems, specially in camera networks with non-overlapping fields of view. A practical re-identification system requires several components: person detection, feature extraction, classification and finally tracking across cameras. For the evaluation and deployment of the algorithms, suitable datasets, evaluation metrics and data presentation formats are needed.
Essay due? We'll write it for you!
In this work the re-identification problem is addressed in many perspectives. We propose novel methods for:
(i) dealing with failures and errors in detection;
(ii) feature extraction using semantic body part segmentation;
(iii) classification using Multi-View optimization techniques;
(iv) temporal integration by window-based classifiers;
(v) evaluation and data presentation for automated systems;
(vi) inter-camera tracking using a Multiple Hypothesis Tracker. The presented methodologies are evaluated in several datasets, including a novel high-definition dataset developed in-house, with applications to re-identification in camera networks.
With the aim of fully automating the re-identification procedure, it was proposed the integration of pedestrian detection methods with the classification stage of re-identification, and an evaluation of the issues arising from that integration was performed. In particular a false positive class was trained to tackle the false positives arising from the detection stage. For feature extraction, the effect of detecting and dividing the human body in semantically valid parts, such as dividing by the waist, or legs, torso and head, was evaluated.
Extracting features from these local regions produces richer descriptors of person’s appearance and increases recognition results consistently. For classification, a Multi-View semi-supervised optimization formulation was used, which integrates in a principled way several features (called views). The stated formulation allows for an optimal closed form solution which assures a fast learning. The semi-supervised aspect of the algorithm is well suited to the reidentification problem, where typically there are few labeled samples and a large number of unlabeled samples. To enhance performance of any singleframe classifier, a window-based wrapper for the classifier was proposed, that filters classification results according to the temporal coherence of pedestrian appearances.
Finally, for inter-camera tracking the Multiple Hypothesis Tracker was used that keeps in memory multiple probable states of the world, which allows the tracker to update its belief based on both past and new information, being able to actually correct previous tracking association mistakes.
Disclaimer: This essay has been submitted by a student. This is not an example of the work written by our professional essay writers. You can order our professional work here.
Sorry, copying is not allowed on our website. If you’d like this or any other sample, we’ll happily email it to you.
Your essay sample has been sent.
Want us to write one just for you? We can custom edit this essay into an original, 100% plagiarism free essay.Order now