Interpreting deep learning models for explainable animal detection using Grad-Cam with RetinaNet, SSD and YOLOv8 | Synapse