Livestock quantity monitoring method and device, computer equipment and storage medium

Abstract

The embodiment of the invention belongs to the technical field of artificial intelligence, and relates to a livestock quantity monitoring method, which comprises the steps of performing livestock detection on an acquired picture according to a target detection AI model to obtain a detection frame; intercepting an image of a single livestock in the detection frame to obtain a livestock sample graph, and processing the livestock sample graph through a depth tracking model and a livestock classification model to obtain a tracking detection matching frame, a redundant tracking frame and a trackingcompensation missing detection frame; adding the number of the obtained tracking compensation leak detection boxes and the number of the tracking detection matching boxes to obtain an actual livestock number as a statistical result; comparing the statistical result with the livestock quantity stored in a database; and if the statistical result is inconsistent with the livestock quantity stored inthe database, judging that the livestock quantity is abnormal, and sending an abnormal alarm to the user. The invention further provides a livestock quantity monitoring method, computer equipment anda storage medium. According to the invention, the multi-detection rate and omission ratio of livestock are effectively reduced.

Description

technical field [0001] The present application relates to the technical field of artificial intelligence, in particular to methods, devices, computer equipment and storage media for monitoring the number of livestock. Background technique [0002] The statistics of the number of livestock in the traditional farms are completely dependent on manual work. Manually go to the farm at regular intervals to record the number of livestock in each pen. However, manual calculations are prone to calculation errors, resulting in situations where the numbers are difficult to correspond to; and when it is found that When the number of livestock is abnormal, it is difficult for breeders to find out the cause in time, resulting in the loss of livestock. [0003] At present, modern farms introduce automation equipment and intelligent software algorithms to improve breeding efficiency and improve breeding quality. Most of the existing methods use a combination of cameras and a single model to...

AI Technical Summary

Problems solved by technology

[0002] The statistics of the number of livestock in the traditional farms are completely dependent on manual work. Manually go to the farm at regular intervals to record the number of livestock in each pen. However, manual calculations are prone to calculation errors, resulting in situations where the numbers are difficult to correspond to; and when it is found that When the number of livestock is abnormal, it is difficult for the breeders to find out the reason in time, resulting in the loss of livestock

[0003] At present, modern farms introduce automation equipment and intelligent software algorithms to improve breeding efficiency and improve breeding quality. Most of the existing methods use a combination of cameras and a single model to count the number of livestock in the farm and monitor the livestock. During the process, non-livestock objects may be mistakenly identified as livestock, resulting in the problem of multi-detection rate

However, in the process of model statistics on the number of livestock, due to the different models used, the situation of missed detection of livestock occurred

Images