Author(s): Saleema N P, Meera K

Abstract: By using a deep convolutional neural network, ie, heavy network architecture we can produce impressive accuracy in many applications. Taking a well-trained heavy network architecture as a guiding module or teache network, we can train a student network that is lightweight yet accurate. In this way, we can make a mobile or portable student network that is accurate as of that of a teacher module. This paper proposes a method to classify multi-object images by using a student-teacher network. This model includes a fully convolutional localization architecture to localize the regions that may contain multiple highly dependent labels. The localized regions are further sent to the recurrent neural networks (RNN) to characterize the latent semantic dependencies at the regional level. Experimental results on several benchmark datasets show that our proposed model achieves the best performance compared to the state-of-the-art models, especially for predicting small objects included in the images.