And Xuliang Duan 1, College of Details Engineering, Sichuan Agricultural University, Ya’an 625000, China; linbin203279@gmail (B.L.); ameter.above.thesky@gmail (Z.X.); [email protected] (F.L.); [email protected] (J.L.); [email protected] (C.M.); [email protected] (X.G.) College of Science, Sichuan Agricultural University, Ya’an 625000, China; [email protected] Correspondence: [email protected]; Tel.: 86-150-083-053-Abstract: A video-based process to quantify animal posture movement is often a potent approach to analyze animal behavior. Each humans and fish can judge the physiological state via the skeleton framework. Nonetheless, it truly is challenging for farmers to judge the breeding state in the complex underwater environment. Consequently, images is often transmitted by the underwater camera and monitored by a laptop or computer vision model. Having said that, it lacks datasets in artificial intelligence and is unable to train deep neural networks. The key contributions of this paper include things like: (1) the world’s initially fish posture database is established. 10 key AUTEN-99 Purity & Documentation points of each fish are manually marked. The fish flock images had been taken inside the experimental tank and 1000 single fish photos had been separated from the fish flock. (2) A two-stage attitude estimation model is employed to detect fish key points. The evaluation of the algorithm overall performance indicates the precision of detection reaches 90.61 , F1-score reaches 90 , and Fps also reaches 23.26. We made a preliminary Aligeron site exploration on the pose estimation of fish and provided a feasible thought for fish pose estimation. Search phrases: aquaculture automation; rotating box; fish detection; fish pose; pc visionCitation: Lin, B.; Jiang, K.; Xu, Z.; Li, F.; Li, J.; Mou, C.; Gong, X.; Duan, X. Feasibility Study on Fish Pose Estimation Primarily based on Rotating Box Object Detection. Fishes 2021, 6, 65. ten.3390/ fishes6040065 Received: 24 October 2021 Accepted: 17 November 2021 Published: 19 November1. Introduction Fish generally have higher nutritional value and may meet the needs of humans and other species. With the improvement of social levels, folks put forward greater and higher specifications for the meat good quality and taste of fish. To meet these higher needs, farmers need to accurately breed and monitor fish in real-time and accurately grasp the distribution, growth status, and behavioral traits of fish [1]. Because of the complicated underwater environment, the adaptability of traditional and backward electronic gear in water is extremely low, as well as dangerous substances can be developed, which interfere with all the living environment of fish, affect their growth, alter their physiological properties, and bring losses in breeding and sales [2]. Consequently, the realization of fishery intelligent detection by a computer vision method will be the inevitable trend on the development of the fishery breeding business chain in modern society. Object detection and pose estimation are essential supporting technologies for fish distribution and condition observation and measurement [3]. Each object detection and pose estimation belong to the standard tasks of machine vision. The former is applied to detect regardless of whether there are target objects of a offered category inside a given image, as well as the latter is utilised to predict the pose from the target object (human or animal) in the input image [4]. As a branch technology of computer vision and image processing, object detection is utilised to detect distinct semantic objects (such.

By mPEGS 1