Bird breed classification and annotation using saliency based graphical model

• We construct a model considering local context, global context and saliency.
• A new method of over-segmentation is proposed.
• We use the feature on the object level to improve the precise.

Due to the variations among the birds, bird breed classification is still a challenging task. In this paper, we propose a saliency based graphical model (GMS), which can precisely annotate the object on the pixel level. In the proposed method, we first over-segment the image into several regions. Then, GMS extracts the object and classifies the image based on the local context, global context and saliency of each region. In order to achieve a high precision of classification, we use SVM to classify the image based on the features of the annotated bird. Finally, we employ posterior probability distribution obtained by GMS and SVM to perform the image classification. Experiments on the Caltech-UCSD Birds dataset show that the proposed model can achieve better results compared with existing bird breed classification methods based on graphical model.