An Intelligent Pig Breeding Surveillance System Based on Video Activity Analysis

Abstract

The invention discloses an intelligent pig captive monitoring system based on video activity analysis, aiming at the problem that the instance-level monitoring of animal behavior in the animal husbandry captive breeding scene still needs to be improved. The invention uses semantic segmentation based on deep convolutional neural network to obtain the mask of the animal; uses the formula based on color saliency to segment the colored ear tag of the animal; the mask for semantic segmentation is processed by the distance transformation method to obtain the instance segmentation result; according to the example As a result of the segmentation, use the distance priority method and posture constraints to associate the ear tag with the instance to obtain the animal skeleton model; based on the animal skeleton model, obtain the attributes of its center point position and head orientation, and add the region of interest counter; based on the interest The counter obtains the tail biting judgment method of the pig. The invention uses surveillance video information, combined with ear tag information, to mark time points that need attention in the video through interest region events based on the skeletal model.

Description

technical field [0001] The invention relates to a computer vision monitoring and analysis system in the scene of animal breeding and captive breeding, in particular to an intelligent pig captive breeding monitoring system based on video activity analysis. Background technique [0002] Animal husbandry monitoring is an industry that responds to the concept of big data and artificial intelligence. In recent years, many sheds for captive animals have deployed video surveillance, but its role is limited to security and manual observation of animal habits. The task characteristics of observing the special behavior of animals determine that the work efficiency of manual observation of animal habits is low. On the one hand, manpower and time are wasted, and on the other hand, there is the possibility of missed detection. These factors further reduce the actual value of video surveillance. The reason for the above-mentioned contradiction is that the video sequence is not introduced ...

AI Technical Summary

Problems solved by technology

[0005] The present invention overcomes the problem in the prior art that the instance-level monitoring of animal behavior in the animal husbandry breeding scene still needs to be improved, and provides an intelligent system based on video activity analysis that can assist manual playback of video and provide reference data for computer analysis. Pig Breeding Monitoring System

Images