Aiming at the problem of performance degradation caused by ignoring the softmax score of the wrong class in the training process of the unbalanced data set of Thangka images and the problem of the loss of negative feature information in the propagation process of the ReLU activation function, a new loss calculation method is proposed. Firstly, the parameters of the pre-training model on the COCO data set are used as the initial parameters. Secondly, CE and CCE are used to calculate the loss during the calculation of loss in the back propagation. Finally, AReLU activation function is used and a weight assigned to CE and CCE is added as final loss to update the parameters. The experimental results show that this algorithm improves the convergence speed and accuracy of the model with respect to imbalanced data. Compared with other loss functions, ours method performance is state-of-the-art, such as complement cross entropy.

No abstract

Thangka is an important part of Tibetan culture, and the classification of Thangka image is one of the basic works of Thangka research. DenseNet(Densely connected convolutional networks) has achieved a very good effect in the field of image classification. Considering that the DenseNet adopts ReLU function which loses the negative feature of the image in the feature propagation process, this paper proposed an improved DenseNet, called L-DenseNet that Leaky ReLU replaces ReLU function to increase the negative feature of propagation. In order to solve the problem of insufficient Thangka image sample, we adopted the method of based fine-tuned network. Experimental results show that L-DenseNet obtains an outstanding performance, which improves 1.1% performance compared with DenseNet. Compared with other CNNs, such as VGG16, ResNet50 and InceptionV3, L-DenseNet obtains state-of-the-art performance in the classification of Thangka image.

No abstract

No abstract