AI Paper Atlas

Research map / Video / Video-language understanding

Video2vec Embeddings Recognize Events When Examples Are Scarce

Amirhossein Habibian, Thomas Mensink, Cees G. M. Snoek

TPAMI 2016 · Video › Video-language understanding · approach: Text-video retrieval

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 VideoStory embedding instead of low-level featuresevent classifier · MED 2013mAP: 28.1 -> 32.4Improved
use VideoStory embedding instead of ImageNet attributesevent classifier · MED 2014mAP: 17.2 -> 24.2Improved
use VideoStory embedding instead of term attributes (no embedding) (ablation)few-example recognition · MED 2013, 2014, CCV10mAPImproved
use VideoStory instead of CCA or description embedding (ablation)embedding · MED 2013, 2014, CCV10mAPImproved
add motion then audio modalities to VideoStoryF (ablation)VideoStoryF · CCV10mAP: 54.8 -> 58.2 -> 59.3Improved
learn embedding jointly over modalities (VideoStoryF) instead of late fusion (ablation)VideoStory · MED 2013mAP: 33.6 -> 37.1Improved
learn embedding jointly over modalities (VideoStoryF) instead of early fusion on low-level features (ablation)VideoStory · CCV10mAP: 56.3 -> 59.3Improved
add term sensitive descriptiveness loss (ablation)VideoStory0 · MED 2013zero-example mAP: 15.9 -> 18.3Improved
add multimodal predictability loss on top of term sensitive loss (ablation)VideoStoryF0 · MED 2013zero-example mAP: 18.3 -> 20.0Improved
add multimodal predictability loss on top of term sensitive loss (ablation)VideoStoryF0 · MED 2014zero-example mAP: 6.8 -> 8.0Improved
use plain VideoStory (LSI loss) instead of term attributes-f for zero examples (ablation)VideoStory · MED 2014zero-example mAP: 5.0 -> 5.2No clear effect
increase positive training examples from 10 to 100 (ablation)VideoStory vs low-level · MED 2013, CCV10mAPMixed

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