Automatic Monitoring Of Bee Colony At Hive Gate Area Based On Computer Vision

Bee industry plays an important role in modern agriculture and is an important part of human living environment and economy.Bee breeding is the foundation and key link of bee industry.Traditional breeding methods rely on manpower and experience,and open hive observation of bees in the process of operation will greatly interfere with the activities and production of bees.With the development of sensors and other technologies,the use of machine vision technology to carry out automatic monitoring and behavior analysis of bees will effectively solve the problem of non-contact observation of bees.Under this background,this paper takes computer vision as the core technology to carry out the research on the automatic monitoring of bee colonies in the hive gate area.Realized the statistics of the number of bee colonies in the hive gate area and the number of bees entering and leaving the hive.Developed the automatic monitoring system of bee colonies in the hive gate area.The research results of this paper provide theoretical basis and technical support for the automatic monitoring of bee colony in the hive gate area,and also lay the data and technical research foundation for the analysis and research of bees' behavior.The main research contents of this paper include the following three parts.(1)The automatic acquisition of the monitoring index of the colony number in the hive gate area is realized.In different light,weather,colony number and other environments,automatic detection statistics are carried out for the colony number in the hive gate area,and algorithm research such as optimized codebook detection,traditional method combining average pixel statistics and deep learning method based on SSD model target detection are carried out.The results show that SSD method has more advantages in detecting bees in different scenes with complex conditions such as bee shadow and bee occlusion.Under the conditions of small,medium and large number of bees,the statistical accuracy rate of the number of bees using SSD method is 96.34%,92.52% and 88.06%,while the accuracy rate of traditional method is only 80.13%,68.30% and 61.01%.SSD method detection results provide an effective solution for colony out-of-hive monitoring and also lay a good detection foundation for further development of bee tracking.(2)The automatic acquisition of monitoring indexes of bees entering and leaving the hive gate area is realized.On the one hand,the research work of bee tracking algorithm is carried out.According to the results of bee detection,sort and deepsort methods are used to track the entering and leaving hive of bees,and an angle-based entering and leaving state judgment method is proposed to analyze the difference between bees entering and leaving the hive.According to the characteristics of bee tracking trajectory and bees state judgment,a statistical method for bees entering and leaving hive is proposed.On the other hand,a statistical study on the number of bees entering and leaving the hive was carried out.The videos of the hive gate area in 30 different environments were used as research objects.The tracking method based on deepsort was used to improve the ID jump problem of bees in the tracking process.Meanwhile,the statistical results showed that the average MSE of bees entering and leaving the hive gate were 95.33 and 69.33 respectively,and the bees tracking results were stable under the conditions of small number of bees,medium number of bees and large number of bees.The automatic monitoring of the number of bees entering and leaving the hive provides an effective index for the monitoring outside the hive,and also provides technical support for the statistical research of bees entering and leaving the hive.(3)An automatic monitoring system for bees in the hive gate area was established.The honeybee detection algorithm and the honeybee tracking algorithm are applied to actual production,and on this basis,an automatic monitoring system for honeybee colonies in the hive gate area is constructed to realize automatic collection of the number of honeybees entering and leaving the hive.The system is developed using a three-tier structure based on B/S.The development platform is eclipse and the database is mysql.The system functions include system management,bee farm management,hive management and other functions.The system realizes automatic monitoring of the number of bees entering and leaving the hive gate area and other indicators,and can build a user early warning mechanism according to the monitoring indicators.