Livestock Fish Larvae Counting using DETR and YOLO based Deep Networks
This work addresses the time-consuming task of fish larvae counting for aquaculture, but it is incremental as it applies existing methods to a new dataset.
The paper tackled the problem of counting fish larvae in aquaculture by evaluating four neural network architectures, including DETR and YOLO variants, on a new annotated dataset, achieving a MAPE of 4.46% with an extra-large DETR and 4.71% with a medium-sized YOLOv8.
Counting fish larvae is an important, yet demanding and time consuming, task in aquaculture. In order to address this problem, in this work, we evaluate four neural network architectures, including convolutional neural networks and transformers, in different sizes, in the task of fish larvae counting. For the evaluation, we present a new annotated image dataset with less data collection requirements than preceding works, with images of spotted sorubim and dourado larvae. By using image tiling techniques, we achieve a MAPE of 4.46% ($\pm 4.70$) with an extra large real time detection transformer, and 4.71% ($\pm 4.98$) with a medium-sized YOLOv8.