Object Detection in Natural Images.
Independent thesis Advanced level (degree of Master (Two Years)), 20 credits / 30 HE creditsStudent thesis
This thesis work is aiming for detecting and localizing different classes of objects in natural images. Algorithm which start from object detection with Implicit Shape model (ISM), then with patch based class segmentation verifies hypothesis with MDL principle are used. This algorithm is able to achieve a high detection rate on both rigid and articulated objects. The thesis also proposes a novel multi-class ISM based on training joint codebook which benefits from sharing features. Experiments have verified that multi-class ISM can reach the same level of detection accuracy as the original ISM while being considerably more efficient. To further improve the computation efficiency, Information Gain(IG) integrated with word frequency criteria has been combined to select codebook entries. Evaluation shows that decreasing 60% of the codebook size would only lead to less than 5% loss in detection accuracy. To further improve the performance, saliency detection can be applied on test images on certain datasets. This has also been shown by the experiments.
Place, publisher, year, edition, pages
Trita-CSC-E, ISSN 1653-5715 ; 2011:112
IdentifiersURN: urn:nbn:se:kth:diva-130795OAI: oai:DiVA.org:kth-130795DiVA: diva2:654242
Master of Science - Systems, Control and Robotics