Simpsons Character X-Ray
A team project for recognizing Simpsons characters in images and video, combining classification, YOLO detection, an X-Ray pipeline, and explainability experiments.
Purpose
This team project brings together several deep-learning workflows for recognising Simpsons characters in images and video. Classification, object detection, video processing, and explainability experiments are studied as connected parts of the problem rather than one single model.
Models and video
The repository includes TensorFlow/Keras classifiers, custom CNN and DINOv2 experiments in PyTorch, and a YOLO detection pipeline. Tools generate classification datasets from labelled boxes. The X-Ray workflow combines detections with character classification across video frames; notebooks examine approaches and evaluation.
XAI and limits
Explainable-AI notebooks investigate which image regions relate to predictions. The repository collects a team project’s code and experiments; listed architectures do not guarantee a particular accuracy or production readiness. Results need to be interpreted by dataset and experiment.