Farm target automatic tracking and monitoring video acquisition system and method

Abstract

The invention relates to a farm target automatic tracking and monitoring video acquisition system and method. The system comprises an inspection robot and an upper computer monitoring system. A trackof the inspection robot is installed above a farm colony house, square limiting devices are arranged at the two ends of the inspection track, the mechanical arm is installed on the inspection track through a connecting rod, a camera is fixed to the other end of the mechanical arm, and the upper computer monitoring system controls the inspection robot. The method comprises the following steps: based on a breeding target detection and positioning result, controlling a mechanical arm to move, performing directional video acquisition, judging abnormal behaviors of a breeding target by an upper computer monitoring system or breeding personnel, determining a suspected abnormal behavior position, controlling a robot to move to a specified position, and performing fixed-point video acquisition. When an improved SSD detection method and an FDSSD detection method are adopted to detect that the breeding targets are excessively gathered or fight, the video collection time is prolonged, and timingvideo collection is carried out; the whole system and method are reasonable in design, easy and convenient to operate and capable of achieving real-time and automatic monitoring of breeding targets.

Description

technical field [0001] The invention belongs to the field of automatic tracking and monitoring of farm video, in particular to a system and method for automatically tracking and monitoring video of a farm target. Background technique [0002] my country is the largest livestock and poultry breeding and consuming country in the world. The total production of livestock and poultry meat accounts for half of the world's total meat. The traditional livestock and poultry breeding is inefficient and cannot meet the growing market demand. From the free-range farming mode of small and medium-sized households to the large-scale and intensive farming mode; in intensive farming, the traditional abnormal monitoring method of breeding targets relies on a large amount of manpower and material resources, which is inefficient and difficult to guarantee the effect. Automatic and intelligent monitoring is the future development trend ; Applying machine vision to livestock and poultry farming can...

AI Technical Summary

Problems solved by technology

[0002] my country is the largest livestock and poultry breeding and consuming country in the world. The total production of livestock and poultry meat accounts for half of the world's total meat. The traditional livestock and poultry breeding is inefficient and cannot meet the growing market demand. From the free-range farming mode of small and medium-sized households to the large-scale and intensive farming mode; in intensive farming, the traditional abnormal monitoring method of breeding targets relies on a large amount of manpower and material resources, which is inefficient and difficult to guarantee the effect. Automatic and intelligent monitoring is the future development trend ; Applying machine vision to livestock and poultry breeding can obtain real-time information on the behavior of breeding targets without interfering with their normal life

[0003] However, in the process of real-time acquisition of breeding target video in the farm, due to the limitation of camera shooting angle and shooting range, there will be visual blind spots, which will affect the tracking and monitoring of breeding targets. In order to automatically monitor breeding targets and achieve full coverage of breeding target videos, traditional methods The monitoring method requires the installation of a large number of cameras, which seriously affects the economy of breeding target tracking and monitoring, and also brings difficulties to the analysis and processing of multi-source video data. A similar system is used in indoor indicator light recognition by Zhejiang Guozi Company. The inspection robot uses the method of converting the collected images into the HSV color space type to calculate the average brightness of the indicator light area in each target image, and count the number of times the indicator lights are on within the preset collection time period. It can process a limited number of images in a period of time, and the real-time performance is poor; Panzhihua University proposed a video detection technology that integrates Gaussian model and Canny operator for traffic information collection. When Canny operator performs edge extraction, the edge in the image It can only be marked once, and the possible image noise should not be marked as an edge, and it cannot be automatically tracked and shot; Tianjin Ya'an Technology Co., Ltd. proposes a video surveillance system for target detection and tracking based on a multi-feature deep neural network. The YOLO framework based on the deep learning algorithm pre-trained the deep learning model, which cannot automatically judge and select the target to be tracked

Images