Sheep individual identity recognition method based on deep learning algorithm

Abstract

The invention discloses a sheep individual identity recognition method based on a deep learning algorithm, and the method comprises the steps: carrying out the bounding box detection of a data set of a sheep individual through employing an SSD network pre-training model, and carrying out the cutting processing of a bounding box; adopting a pre-training model Resnet18 of transfer learning to learn individual features of the sheep; carrying out identity recognition on the individual sheep by adopting a joint optimization loss function of a triple loss function and a cross entropy loss function; wherein the optimal coefficient of the joint optimization loss function is determined by adopting an experimental comparison method. The invention can automatically and efficiently carry out image-based intelligent recognition on the individual sheep, thereby saving the time cost and the marking cost of manual marking of professional breeding personnel, not hurting the individual sheep according to the principles of no contact, low cost, high yield and no harm, and providing a deep learning thought of identification of the individual sheep at the same time. And a research thought is also provided for identification of animal individual identities with high similarity.

Description

technical field [0001] The present invention relates to the technical field of computer graphics, and more specifically to a sheep individual identification method based on a deep learning algorithm. Background technique [0002] In the modern breeding and breeding work, animal husbandry is a vital branch of agricultural development, especially for Northwest China, improving the intelligent development of animal husbandry is the key to improving the rural economy and agricultural living standards. Real-time detection and identification of individuals in smart breeding is the primary problem that needs to be solved urgently. By grasping the real-time information of individual animals or groups, emergencies can be warned; and the health status of individual animals can be grasped in real time by constructing different evaluation indicators. At the same time, individual identification is also one of the basic technologies for realizing big data management and breeding. [0003...

AI Technical Summary

Problems solved by technology

The traditional methods of marking animals, whether it is ear branding or external ear tags, are likely to cause varying degrees of damage to animals. For example, the traditional branding method is very irritating to animals, which will cause certain harm to livestock In addition, the ear tag external tag method will not only cause the animal's ear to be easily infected, but also cause the loss of the ear tag due to the activity of the animal itself.

Images