AI Paper Atlas

Research map / Vision-language and multimodal / CLIP and zero-shot

MAtch, eXpand and Improve: Unsupervised Finetuning for Zero-Shot Action Recognition with Language Knowledge

Wei Lin, Leonid Karlinsky, Nina Shvetsova, Horst Possegger, Mateusz Kozinski, Rameswar Panda, Rogerio Feris, Hilde Kuehne, Horst Bischof

ICCV 2023 · Vision-language and multimodal › CLIP and zero-shot · approach: Adapting VLMs for zero-shot classification · 47 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
finetune CLIP on unlabeled K400 with MAXI (K400 dict only)CLIP ViT-B/16 · HMDB51 zero-shottop-1 accuracy: 38.0 -> 50.5Improved
add GPT-3 verbs to the text bag (ablation)MAXI · UCF101 zero-shottop-1 accuracy: 76.6 -> 77.8Improved
add BLIP verbs on top of GPT-3 verbs (ablation)MAXI · UCF101 zero-shottop-1 accuracy: 76.45 -> 77.88Improved
use BLIP verbs instead of BLIP object nouns (ablation)MAXI text bag · HMDB51 zero-shottop-1 accuracy: 50.15 -> 50.92Improved
filter text bags to top 90% by CLIP similarity (ablation)MAXI · HMDB51 zero-shottop-1 accuracy: 51.09 -> 52.24Improved
aggressive filtering to 30% of videos (ablation)MAXI · HMDB51 zero-shottop-1 accuracy: 51.09 -> 47.73Hurt
use MIL-NCE instead of cross entropy over fixed classes (ablation)MAXI · K600 zero-shottop-1 accuracy: 65.09 -> 71.24Improved
add same-best-matched-text positives to MIL-NCE (ablation)MAXI · K600 zero-shottop-1 accuracy: 70.14 -> 71.24Improved
increase text bag size from 1 to 16 (ablation)MAXI · HMDB51 zero-shottop-1 accuracy: 49.85 -> 51.09Improved
use MAXI instead of CLIP initialization for few-shot (2-shot UCF101)ViFi-CLIP few-shot · UCF101 2-shotaccuracy: 84.0 -> 86.8Improved
use under-specified (200-class) action dictionary (ablation)MAXI · UCF101 zero-shottop-1 accuracy: 78.18 -> 75.10Hurt
use over-specified dictionary with 400 random added verbs (ablation)MAXI · HMDB51 zero-shottop-1 accuracy: 50.35 -> 45.97Hurt

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