Core Concepts
Opti-CAM optimizes saliency maps for interpretability in deep neural networks by combining CAM-based and masking-based approaches.
Stats
In this work we introduce Opti-CAM, combining ideas from CAM-based and masking-based approaches.
Opti-CAM largely outperforms other CAM-based approaches according to the most relevant classification metrics.
On several datasets, Opti-CAM improves the state of the art by a large margin.
Quotes
"Opti-CAM combines ideas from CAM-based and masking-based approaches."
"Opti-CAM largely outperforms other CAM-based approaches."
"Opti-CAM improves the state of the art by a large margin."