Large Scale Visual Recognition through Adaptation using Joint Representation and Multiple Instance Learning
Journal of Machine Learning Research, Volume 17, 2016, Pages 142:1-142:31.
A major barrier towards scaling visual recognition systems is the difficulty of obtaining labeled images for large numbers of categories. Recently, deep convolutional neural networks (CNNs) trained used 1.2M+ labeled images have emerged as clear winners on object classification benchmarks. Unfortunately, only a small fraction of those lab...More
Full Text (Upload PDF)
PPT (Upload PPT)