The attractivity assessment of raw materials on Penaeus vannamei is carried out to evaluate their potential to stimulate feeding behavior and improve feed intake. This helps optimize feed formulation. To achieve this, an artifical intelligence tool was developped in order to track and analyze P. vannamei movement facing different African raw materials.
As a first step, five shrimp are placed in an attractivity measurement tank above which a camera is placed to record them during 5 minutes (figure 1).
The four different raw materials studied are : fermented soy, soy beans, soy cake and toasted cowpea.
The different variables analyzed are : latency time, total distance and speed until reaching the feeding are, number of times the shrimp entered the feeding area and time spent in the feeding area, number of shrimp in the feeding area, time spent in movement, exploration percentage.
The different results obtained show that 67% of the shrimp entered the feeding area for fermented soy and soy cake against 73% for soybeans and toasted cowpea.
No significant differences are observed for the time spent in the feeding area and the number of time shrimp entered the feeding area. Regarding the area, only the speed until reaching it shows a signifiance difference between fermented soy and soybeans with a median value of 3.77 cm/sec and 5.31 cm/sec respectively. This difference is also observed for the time spent moving with 107.29 sec and 184.76 sec respectively. Heatmaps are produced per video for an individual or the whole group.
The Artificial Intelligence tool shows a good capacity of detecting and tracking shrimp movement in the attractivity measurement tank despite the complexity of their behavior and the disturbing elements : individuals overlapping, tail flip, disappearance and camera reflection. This performance enhance the assessment of shrimp feeding behavior.