Observation Of Early Development Of European European Seabass And Study On Automatic Identification Of Abnormal Fish

This study consists of three parts.The first part is the observation of embryonic development and larval growth of the European seabass Dicentrarchus labrax;the second part is the starvation experiment and irreversible point study of the Dicentrarchus labrax larvae;The third part is based on computer vision and abnormal swimming behavior of malformed fish identification researchIn order to understand the developmental characteristics of Dicentrarchus labrax embryos and larvae,the artificial fertilization was used to obtain the fertilized eggs.Under the conditions of artificial hatching and cultivation,the embryonic development and morphological changes and growth characteristics of the carp were observed.The study founded that the fertilized eggs were round,buoyant and had an egg diameter of 1.287±0.025 mm.At 13±0.5 °C water temperature,embryonic development had gone through the stages of placenta formation,cleavage stage,blastocyst stage,gastrula stage,neuroblast stage,organ formation stage,etc.,hatching out the membrane for 72 h,The total length of the first hatching larvae was 3.98 ± 0.02 mm.At 13±0.5 °C water temperature,0-9 d was in the early stage of larvae: 0-5 days of hatching,the larvae swimming ability was poor,the heart,intestines,fins and other organs continued to develop,and the yolk sac accounted for a large proportion of the body;Five days after incubation,the color of the crystals in the eye capsules became black,the melanin on the surface gathered,and the yolk sac became significantly smaller;the larvae,the mouth cracks and the pectoral fins formed on the 7th day after hatching,a few larvae began to swim,but could not feeding,still rely on the yolk.The capsule provided nutrition.After 9 days,it entered the late stage of larvae: almost all of the larvae which hatched for 9 days entered the parallel state,food was visible in the intestines,and the yolk sac did not completely disappear,which was a mixed vegetative period;the yolk sac of the larvae that hatched for 11 days completely disappeared,and the battalion was exogenous nutrition;the larvae swimming ability increased from 13 to 17 d,the intestines were full of food,and entered the rapid growth period;the larvae hatched for 19-19 days,the individual began to show significant size difference;after 19 days of hatching,the fish became black and the fins basically formed.Continued development entered the juvenile period.The mixed vegetative period of the oyster larvaewas short,and the growth equation showed the allometric growth characteristics.This study accumulated the early life history data of Dicentrarchus labrax,and provided a theoretical reference for artificial propagation and fry breeding of oyster.The effects of starvation on the growth,morphology and survival of larvae were studied by experimental ecological methods.The initial feeding rate and irreversible point(PNR)of starved larvae were determined.The results showed that the initial feeding rate of Dicentrarchus labrax larvae was only 12%,and the individual developmental speed of juveniles was different.The irreversible growth point of eucalyptus was the 10 th day after hatching;the growth rate of hunger group after opening was significantly lower than that of the control group,yolk sac and oil ball disappeared faster.Hunger can caused larval developmental malformations.The study identified irreversible hunger and explored the effects of hunger on larval development.It provided a new basis for judging the degree of hunger and management in fry production.To solve the problem of automatic identification of deformed fish.Based on computer vision technology and abnormal swimming behavior,this paper studied the identification of the scorpion fish.Firstly,the motion of the fish was extracted by the motion influence map(MIM),and the abnormal swimming behavior was initially positioned and then taken into the screenshot window.The YOLO technique was used to zoom the window of the positioned individual.The window contained only one fish individual;the DCGANs algorithm was used to classify the pictures to determine whether they were deformed fish or normal fish.In the case of malformed fish,the type of deformed fish was further distinguished.The results showed that the average recognition rate of malformed fish can reached more than 90%,and the recognition accuracy of two different malformation types can reached more than 80%.This method can greatly eliminated the dependence on artificially selected malformed fish,reduced the cost,and provided a non-destructive,economical and effective way for fish behavior recognition.