Investigation and analysis of image classification on large-scale benchmark datasets
Metadata[+] Show full item record
[ACCESS RESTRICTED TO THE UNIVERSITY OF MISSOURI AT REQUEST OF AUTHOR.] Image Classification nowadays, which including object recognition and scene classification, remains to be a major challenging task among computer recognition area. Defined as the task of assigning an image one or multiple labels corresponding to the presence of a category in the image, the difficulties of image classification results from intra-class variations, viewpoint changes and deformations of the objects, etc. In the thesis, first, an overview of a series of the state-of-the-art image classification frameworks will be introduced, such as the most popularized bag-of-words method, the spatial pyramid matching algorithm and the convolutional neural networks; Then an in-depth view of the image classification challenges will be discussed; Last but mot the least, the experiments and the experimental results regarding to the proposed feature transfer algorithm suited for image classification on large-scale data-sets such as PASCAL VOC and ImageNet will be talked about as well.
Access to files is limited to the University of Missouri--Columbia.