Alert button

TempSAL -- Uncovering Temporal Information for Deep Saliency Prediction

Jan 05, 2023
Figure 1 for TempSAL -- Uncovering Temporal Information for Deep Saliency Prediction
Figure 2 for TempSAL -- Uncovering Temporal Information for Deep Saliency Prediction
Figure 3 for TempSAL -- Uncovering Temporal Information for Deep Saliency Prediction
Figure 4 for TempSAL -- Uncovering Temporal Information for Deep Saliency Prediction

Share this with someone who'll enjoy it:

Deep saliency prediction algorithms complement the object recognition features, they typically rely on additional information, such as scene context, semantic relationships, gaze direction, and object dissimilarity. However, none of these models consider the temporal nature of gaze shifts during image observation. We introduce a novel saliency prediction model that learns to output saliency maps in sequential time intervals by exploiting human temporal attention patterns. Our approach locally modulates the saliency predictions by combining the learned temporal maps. Our experiments show that our method outperforms the state-of-the-art models, including a multi-duration saliency model, on the SALICON benchmark. Our code will be publicly available on GitHub.

* 10 pages, 7 figures  

Share this with someone who'll enjoy it: