Abstract:Focused on the issue that in imbalanced image classification the recall of minority class is low, the cost of classification is high and feature selection manually costs too much, an imbalanced image classification approach based on convolutional neural network and cost sensitive learning was proposed(Triplet-CSSVM). This method has two parts: feature learning and cost sensitive classification. Firstly, the coding method which mapped images to a Euclidean space end-to-end was learned by the convolution neural network which used triplet loss as loss function(triplet-sampling CNN). At the same time, the dataset was rescaled by sampling method to balance the distribution. Then, the best and the minimum cost classification result was obtained by cost-sensitive support vector machine classification algorithm (CSSVM) which assigned different cost factors to different classes. Experiment with the portrait dataset FaceScrub on the deep learning framework Caffe. And the results show that the recall of the proposed method has up to 93% in the condition of 1: 3 imbalanced rate, and has better performance in precision and F-Score.