AI Paper Atlas

Research map / Video / Video-language understanding

Saliency Prediction in the Deep Learning Era: Successes and Limitations

Ali Borji

TPAMI 2019 · Video › Video-language understanding · approach: Event and anomaly reasoning in video · 242 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
use deep (CNN) saliency models instead of best classic modelsaliency models · MIT300NSS: 1.41 -> 2.47Improved
use deep models instead of best classic modelsaliency models · MIT300AUC-Judd: 0.83 -> 0.88Improved
use deep models instead of best non-deep modelsaliency models · CAT2000AUC-Judd: 0.85 -> 0.88Improved
use deep models instead of best non-deep modelsaliency models · CAT2000NSS: 1.67 -> 2.39Improved
switch evaluation metric from AUC-Judd to AUC-BorjiEML-NET · MIT300rank: 2nd -> 44thMixed
add temporal information to classic video saliency models vs adding deep CNNsvideo saliency models · DHF1KNSS/CC/AUCImproved
use a subset of 10 representative images to rank modelsMIT benchmark · MIT300Spearman correlation: 0.97Improved
replace fixations with mouse data from 90 participantsmouse-click saliency · OSIEfixation prediction accuracyHurt
use BubbleView clicks instead of SALICON mouse movementsclick data · OSIENSS vs fixations: clicks of 10 participants explain about 90% of fixationsImproved
apply image transformations (20 types) at test timedeep saliency models · transformed imagesmodel performanceHurt
augment saliency maps with annotated face/text/person regionsdeep saliency models · MIT300NSSImproved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy