AI Paper Atlas

Research map / Label-efficient and robust learning / Recognition and long-tail

Summarization and Classification of Wearable Camera Streams by Learning the Distributions Over Deep Features of Out-Of-Sample Image Sequences

Alessandro Perina, Sadegh Mohammadi, Nebojsa Jojic, Vittorio Murino

ICCV 2017 · Label-efficient and robust learning › Recognition and long-tail · approach: Loss design and classifier theory · 6 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
add constrained Markov walk (CG -> CGcw), 1 label/class (ablation)counting grid · AIHSaccuracy: 0.2914 -> 0.3144Improved
add constrained Markov walk, 3 labels/class (ablation)counting grid · AIHSaccuracy: 0.3455 -> 0.3627Improved
add constrained Markov walk, 1 label/class (ablation)counting grid · MEAaccuracy: 0.7583 -> 0.7646Improved
add constrained Markov walk, 3 labels/class (ablation)counting grid · MEAaccuracy: 0.801 -> 0.8140Improved
add constrained Markov walk, 1 label/class (ablation)counting grid · Google Glass SetAaccuracy: 0.2001 -> 0.2185Improved
add constrained Markov walk, 3 labels/class (ablation)counting grid · Google Glass SetBaccuracy: 0.3028 -> 0.3217Improved
add constrained Markov walk, full labels, subject A (ablation)counting grid · Google Glass (full labels) subject Aaccuracy: 0.8256 -> 0.8337Improved
add constrained Markov walk, full labels, subject C (ablation)counting grid · Google Glass (full labels) subject Caccuracy: 0.7951 -> 0.8103Improved
add constrained walk for only the first 10% of EM iterations (ablation)counting grid · AIHS 45-classaccuracyImproved
use CGcw embedding instead of fc6 features with NN, 1 label/classfc6 features · AIHSaccuracy: 0.1821 -> 0.3144Improved
use 200D autoencoder instead of CGcw with abundant unlabeled dataCGcw · Google Glass (full labels) subject Aaccuracy: 0.8337 -> 0.9390Hurt
vary grid-window ratio (ablation)counting grid · AIHSaccuracy: around 3% accuracyNo clear effect

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