Cross-Domain Fine-Grained Classification: A Review
Fine-grained classification is an interesting but challenging task due to the high amount of data needed to achieve a high accuracy. However, the high specificity of the classes makes it difficult to collect a large amount of samples. Thus, the use of cross-domain learning is an interesting aspect since an abundant amount of data exists for some domains like web images exists. In this review, the current works of cross-domain fine-grained classification are summarized and potential areas for future work are highlighted. Even though first works exist, the variety of methods is still small and interesting cross-domain settings are rarely considered. Thus, the field of cross-domain fine-grained classification provides a large room for future research.