Search

Monitoring animal behavior can facilitate conservation efforts by providing key insights into wildlife health, population status, and ecosystem function. Automatic recognition of animals and their behaviors is critical for capitalizing on the large unlabeled datasets generated by modern video devices and for accelerating monitoring efforts at scale. However, the development of automated recognition systems is currently hindered by a lack of appropriately labeled datasets. Existing video datasets 1) do not classify animals according to established biological taxonomies; 2) are too small to facilitate large-scale behavioral studies and are often limited to a single species; and 3) do not feature temporally localized annotations and therefore do not facilitate localization of targeted behaviors within longer video sequences. Thus, we propose MammalNet, a new large-scale animal behavior dataset with taxonomy-guided annotations of mammals and their common behaviors. MammalNet contains over 18K videos totaling 539 hours, which is ~10 times larger than the largest existing animal behavior dataset. It covers 17 orders, 69 families, and 173 mammal categories for animal categorization and captures 12 high-level animal behaviors that received focus in previous animal behavior studies. We establish three benchmarks on MammalNet: standard animal and behavior recognition, compositional low-shot animal and behavior recognition, and behavior detection. Our dataset and code have been made available at: https://mammal-net.github.io., Comment: CVPR 2023 proceeding

Successful aquaculture systems can reduce the pressure and help secure the most diverse and productive Red Sea coral reef ecosystem to maintain a healthy and functional ecosystem within a sustainable blue economy. Interestingly, recirculating aquaculture systems are currently emerging in fish farm production practices. On the other hand, monitoring and detecting fish behaviors provide essential information on fish welfare and contribute to an intelligent production in global aquaculture. This work proposes an efficient approach to analyze the spatial distribution status and motion patterns of juvenile clownfish \textit{(Amphiprion bicinctus)} maintained in aquaria at three stocking densities (1, 5, and 10 individuals/aquarium). The estimated displacement is crucial in assessing the dispersion and velocity to express the clownfish's spatial distribution and movement behavior in a recirculating aquaculture system. Indeed, we aim to compute the velocity, magnitude, and turning angle using an optical flow method to assist aquaculturists in efficiently monitoring and identifying fish behavior. We test the system design on a database containing two days of video streams of juvenile clownfish maintained in aquaria. The proposed displacement estimation reveals good performance in measuring clownfish's motion and dispersion characteristics leading to assessing the potential signs of stress behaviors. We demonstrate the effectiveness of the proposed technique for quantifying variation in clownfish activity levels between recordings taken in the morning and afternoon at different stocking densities. It provides practical baseline support for online predicting and monitoring feeding behavior in ornamental fish aquaculture., Comment: 14 pages, 15 figures

This paper studies the fish growth trajectory tracking via reinforcement learning under a representative bioenergetic growth model. Due to the complex aquaculture condition and uncertain environmental factors such as temperature, dissolved oxygen, un-ionized ammonia, and strong nonlinear couplings, including multi-inputs of the fish growth model, the growth trajectory tracking problem can not be efficiently solved by the model-based control approaches in precision aquaculture. To this purpose, we formulate the growth trajectory tracking problem as sampled-data optimal control using discrete state-action pairs Markov decision process. We propose two Q-learning algorithms that learn the optimal control policy from the sampled data of the fish growth trajectories at every stage of the fish life cycle from juveniles to the desired market weight in the aquaculture environment. The Q-learning scheme learns the optimal feeding control policy to fish growth rate cultured in cages and the optimal feeding rate control policy with an optimal temperature profile for the aquaculture fish growth rate in tanks. The simulation results demonstrate that both Q-learning strategies achieve high trajectory tracking performance with less amount feeding rates.

In precision aquaculture, the primary goal is to maximize biomass production while minimizing production costs. This objective can be achieved by optimizing factors that have a strong influence on fish growth, such as the feeding rate, temperature, and dissolved oxygen. This paper provides a comparative study of three model predictive control (MPC) strategies for fish growth reference tracking under a representative bioenergetic growth model in precision aquaculture. We propose to evaluate three candidate MPC formulations for fish growth reference tracking based on the receding horizon. The first MPC formulation tracks a desired fish growth trajectory while penalizing the feed ration, temperature, and dissolved oxygen. The second MPC optimization strategy directly penalizes the feed conversion ratio (FCR), which is the ratio between food quantity and fish weight gain while minimizing the actual growth state's deviation from the given reference growth trajectory. The third MPC formulation includes a tradeoff between the growth rate trajectory tracking, the dynamic energy and the cost of food. Numerical simulations that integrate a realistic bioenergetic fish growth model of Nile tilapia (Oreochromis niloticus) are illustrated to examine the comparative performance of the three proposed optimal control strategies.