Breeding Data Intelligent Acquisition Technology Based On Morphological And Behavioral Parameters Of Prawns And Its Application

In this paper,we carried out prawn morphological and behavioral research based on artificial intelligence(AI)technologies such as machine vision,focusing on technology frontier like intelligent weighing?intelligent measurement and intelligent evaluation of characteristics,invented marine biological weighing software?image measurement technology and intelligent video surveillance cooperating with school of Aerospace engineering?school of Information Science and Engineering of Xiamen University,tested and applied in the practical breeding.The main results are as follows:There were four problems in electronic balance:First was to manually read the data and record it;Second,it was difficult to get accurate data quickly during prawn's jumping;Third,reading was effected by wind;Fourth,we needed to wipe or adjust zero frequently because of ponding.To this end,we cooperated with school of Aerospace engineering and invented marine biological weighing software,which had two advantages:(1)Fast and accurately.The sofeware could finish data capture during prawn's jumping,the speed of not taking out form was 20?23 tails per minute,and it supported to export data;(2)Intelligent recording.The software set up a data intelligence filtering program to avoid false readings,and there was no limit of time interval between the adjacent weighing.The use of marine biological weighing software can accelerate the collection speed of samples in actual breeding,minimize stress and improve the survival rate after weighing,which had practical value.After inventing intelligent weighing technology,this paper further developed intelligent measurement technology for the morphological parameters of prawn.Traditionally people used vernier caliper,but the error was larger and the errors between different people were not consistent.In order to improve the intelligent level of prawns measurement,we cooperated with school of Information Science and Engineering,inventing image measurement technique based on computer vision technology.Using this technique,we measured body length,carapace length and carapace width of Marsupenaeus japonicus,Litopenaeus vannamei and Metapenaeus ensis,then compared these results with that from vernier calipers.It turned out that:(1)More than 95%of the D-value between two methods fell within the scope of LoA,and LoA CI was acceptable in professional sense,which showed that the consistency of two methods was well.(2)For one individual,the results of identifying one angle by using image measurement technique were the same.The ICC of three different angles were 0.996?0.973?0.957,which had no significant difference compared with that of 3 times measurements using vernier caliper by one person(ICC=0.997?0.980?0.965),but larger than 3 times measurements by three person(ICC=0.991?0.952?0.965),Meanwhile,the degree of variation was minimal,the relative error was 1.52%,2.37%and 3.74%respectively.It indicated that the repeatability of image measurement technique was related to shooting angle,the repeatability of the same angle was the best.(3)The larger the shrimp was,the more suitable for image measurement technology.In short,the consistency of image measurement technique and vernier caliper method was well,and the repeatability of the former was superior to the latter.Image measurement technique had advantages in non-contact?fast measurement?wide application and good repeatability,therefore,it could replace vernier caliper to be applied in prawns morphological parameters measurement.In addition to static measurement and weighing of prawn,AI had more important scientific value and application prospect in the evaluation of prawn behavior.There were two kinds of research methods of marine organism's sports,one was behavior observation,the other was video analysis,the former had big workload and its counting time interval was long,also greatly influenced by subjectivity,the latter had been relatively mature in the study of fish.It could provided scientific basis for evaluating aquatic animal behavior and its biological significance.In order to improve the data accuracy and intelligent level of prawn behavioral research,we cooperated with professor and graduate student from school of Information Science and Engineering,inventing intelligent video surveillance,and designed water temperature mutation experiment to test this technology,got conclusions as following:(1)The motion of M.japonicus had no prominent influence in 4? water temperature difference,when the rise of water temperature was larger,the motion of M.japonicus became gentler,the activity frequency were reduced,but the maximum acceleration were increasing.(2)When water temperature increased sharply,there was an 6?11 min adaptation period of M.japonicus,after that M.japonicus gradually stabilized in a certain range;(3)There were two kinds of circumstances of 25 s before reaching the maximum acceleration of M.japonicus:38.67%were basic at rest,mainly appeared in 20?28 0C group;61.33%were in a state of continuous variable motion,20?32 ?and 20?36 ? group were the most;(4)Mjaponicus would turn on one side under large temperature mutation,mainly concentrated in 20?32 ? and 20?36 ? groups,and would be back after adapting to temperature,20?36 ? group needed more time than 20?32 0C group.(5)The application of video analysis technology had greatly reduced the workload,output high accuracy data,and achieved the identification of prawn 's side turn and death,which had applied value in the study of prawn behavior.This paper also used the above techniques to carry on the comprehensive test.Useing marine biological weighing software and image measurement technique to track growth situation of M.japonicus from putting seedling and explored the rule of growth.The results show that:The relationship between body length(L)and body weight(W)was described by the power function:W = 1.15×10~-5 L~3.004.The Von Bertallanffy equations of L and W were expressed as the follows:Lt = 98.28(1-e~-0 022t+0.111)?Wt=11.12(1-e~-0.022t+0.111)3.004,from these growth curves,we found that L had no inflection point and 55d was the growth inflection point of W.Combined with the measured values and the theoretical predictive values calculated by growth equations,the growth of M.japonicus could be divided into three stages:the fast growth stage(20?52 days),the steady growth stage(45?80 days)and the slow growth stage(after 80 days).The above growth pattern was in accordance with the general pattern of prawn's growth,which indicated that it was a successful application of intelligent acquisition technology in practice.