An Improved Convolutional Neural Network Model with Adversarial Net for Multi-label Image Classification

Abstract

Convolution neural network (CNN) achieves outstanding results in single-label image classification task. However, due to the complex underlying object layout and insufficient multi-label training images, it is still an open problem that how CNN better handle multi-label images. In this work, we proposes an improved deep CNN model with Adversarial Net which can extract features of objects at different scales in multi-label images by spatial pyramid pooling. In model training, we first transfer the parameters pre-trained on ImageNet to our model, then train an Adversarial Network to generates examples with occlusions and combine it with our model, which make our model invariant to occlusions. Experimental results on Pascal VOC 2012 multi-label image dataset demonstrate the superiority of the proposed approach over many state-of-the-arts approaches.