Locomotor posture and swimming-intensity quantification in starvation-stress behavior detection of individual fish

Abstract

The welfare of fish is gaining importance in aquaculture. The starvation-stress behavior of fish is important for quantifying their welfare. Therefore, to detect the stress behavior of individual fish under starvation stress, this study proposes swimming intensity as a new parameter for quantifying the swimming activity of fish. This parameter considers the angular information of the fish, including the steering angle, tail-bending angle, and turning speed, as key factors. First, angular features are extracted. A multi-scale cascade perceptual network is proposed to generate limb masks of individual fish. The centroid coordinates of the fish limbs are calculated using the limb mask. The angular features of the fish are calculated by establishing a follower coordinate system using centroid coordinates of the limb. Considering the influence of the occlusion of multiple fish on limb-centroid extraction, two methods are proposed to estimate the centroid coordinates of fish limbs in slight and severe occlusions. Entropy and statistics are used to measure the magnitude of change in the motion of individual fish. Finally, the swimming intensity of the fish is quantified by combining the amplitude of the change in motion and turning speed. Results obtained via human observation and the evacuation of the gastrointestinal content-based long experiments of individual darkbarbel catfish (Pelteobagrus vachelli) reveal that the proposed method exhibits good performance in detecting starvation-stress behavior of darkbarbel catfish with an accuracy rate of at least 96.21 ± 1.42 % and a maximum false negative rate of 0.21 ± 0.05 %. The purpose of the study was to establish a potentially effective method for detecting and recognizing the starvation-stress behavior of individual fish to provide reliable theoretical support for the study of aquaculture management, particularly the precision-feeding strategy of fish, to ensure fish welfare.