Investigation of Animal Image Retrieval Algorithm Based on Deep Learning
With the coming of the big data era, the number of animal images on the Internet is increasing dramatically. Facing with a large amount of animal image data, retrieving the semantics, attributes, annotations, features of an image manually requires a lot of time and labor costs .Due to subjective bias of people, whether it is an marker or a inquirer, cannot describe all the information of the image accurately and consistently, how to retrieve relevant animals quickly from the massive animal data has become an urgent problem to be solved, this brings challenges to the search. The approximate neighbor search method represented by hash has received extensive attention and research. Image retrieval technology is a technology that can quickly and accurately query images .The similarity matching algorithm is used to its similar images which is found in the image library. This paper does some research on image retrieval based on animal content. At present, the focus of animal image retrieval is on how to accurately and completely extract key areas of animal characteristics. Compared with the traditional content-based image retrieval technology, the image retrieval system designed in this paper eliminates the traditional image retrieval system which select image features manually .Can directly input sample data training, network can learn and extract the basic features of the image automatically. Comparison of experimental results shows, image retrieval algorithm based on deep learning has higher retrieval efficiency, it can reach 84.5%, which has higher retrieval efficiency than the classic image retrieval system.