TASK	DATASET	MODEL	METRIC NAME	METRIC VALUE	GLOBAL RANK
Zero-Shot Action Detection	ActivityNet-1.3	UnLoc-B (75% seen split)	mAP IOU@0.5	40.2	# 3
Zero-Shot Action Detection	ActivityNet-1.3	UnLoc-B (50% seen split)	mAP IOU@0.5	36.9	# 6
Zero-Shot Action Detection	ActivityNet-1.3	UnLoc-L (50% seen split)	mAP IOU@0.5	43.2	# 2
Zero-Shot Action Detection	ActivityNet-1.3	UnLoc-L (75% seen split)	mAP IOU@0.5	47.4	# 1
Temporal Action Localization	ActivityNet-1.3	UnLoc-L	mAP IOU@0.5	59.3	# 5
Natural Language Moment Retrieval	ActivityNet Captions	UnLoc-B	R@1,IoU=0.5	48.0	# 4
Natural Language Moment Retrieval	ActivityNet Captions	UnLoc-B	R@1,IoU=0.7	29.7	# 4
Natural Language Moment Retrieval	ActivityNet Captions	UnLoc-B	R@5,IoU=0.5	81.5	# 1
Natural Language Moment Retrieval	ActivityNet Captions	UnLoc-B	R@5,IoU=0.7	61.4	# 2
Natural Language Moment Retrieval	ActivityNet Captions	UnLoc-L	R@1,IoU=0.5	48.3	# 3
Natural Language Moment Retrieval	ActivityNet Captions	UnLoc-L	R@1,IoU=0.7	30.2	# 2
Natural Language Moment Retrieval	ActivityNet Captions	UnLoc-L	R@5,IoU=0.5	79.2	# 3
Natural Language Moment Retrieval	ActivityNet Captions	UnLoc-L	R@5,IoU=0.7	61.3	# 3
Moment Retrieval	Charades-STA	UnLoc-L	R@1 IoU=0.5	60.8	# 4
Moment Retrieval	Charades-STA	UnLoc-L	R@1 IoU=0.7	38.4	# 5
Moment Retrieval	Charades-STA	UnLoc-L	R@5 IoU=0.5	88.2	# 4
Moment Retrieval	Charades-STA	UnLoc-L	R@5 IoU=0.7	61.1	# 2
Moment Retrieval	Charades-STA	UnLoc-B	R@1 IoU=0.5	58.1	# 9
Moment Retrieval	Charades-STA	UnLoc-B	R@1 IoU=0.7	35.4	# 10
Moment Retrieval	Charades-STA	UnLoc-B	R@5 IoU=0.5	87.4	# 5
Moment Retrieval	Charades-STA	UnLoc-B	R@5 IoU=0.7	59.1	# 3
Action Segmentation	COIN	UnLoc-L	Frame accuracy	72.8	# 1
Moment Retrieval	QVHighlights	UnLoc-L	R@1 IoU=0.5	66.1	# 4
Moment Retrieval	QVHighlights	UnLoc-L	R@1 IoU=0.7	46.7	# 12
Moment Retrieval	QVHighlights	UnLoc-B	R@1 IoU=0.5	64.5	# 8
Moment Retrieval	QVHighlights	UnLoc-B	R@1 IoU=0.7	48.8	# 5

Badge	Markdown
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/unloc-a-unified-framework-for-video/zero-shot-action-detection-on-activitynet-1-3)](https://paperswithcode.com/sota/zero-shot-action-detection-on-activitynet-1-3?p=unloc-a-unified-framework-for-video)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/unloc-a-unified-framework-for-video/action-segmentation-on-coin)](https://paperswithcode.com/sota/action-segmentation-on-coin?p=unloc-a-unified-framework-for-video)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/unloc-a-unified-framework-for-video/natural-language-moment-retrieval-on)](https://paperswithcode.com/sota/natural-language-moment-retrieval-on?p=unloc-a-unified-framework-for-video)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/unloc-a-unified-framework-for-video/moment-retrieval-on-charades-sta)](https://paperswithcode.com/sota/moment-retrieval-on-charades-sta?p=unloc-a-unified-framework-for-video)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/unloc-a-unified-framework-for-video/moment-retrieval-on-qvhighlights)](https://paperswithcode.com/sota/moment-retrieval-on-qvhighlights?p=unloc-a-unified-framework-for-video)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/unloc-a-unified-framework-for-video/temporal-action-localization-on-activitynet)](https://paperswithcode.com/sota/temporal-action-localization-on-activitynet?p=unloc-a-unified-framework-for-video)`

UnLoc: A Unified Framework for Video Localization Tasks

ICCV 2023 · Shen Yan, Xuehan Xiong, Arsha Nagrani, Anurag Arnab, Zhonghao Wang, Weina Ge, David Ross, Cordelia Schmid ·

While large-scale image-text pretrained models such as CLIP have been used for multiple video-level tasks on trimmed videos, their use for temporal localization in untrimmed videos is still a relatively unexplored task. We design a new approach for this called UnLoc, which uses pretrained image and text towers, and feeds tokens to a video-text fusion model. The output of the fusion module are then used to construct a feature pyramid in which each level connects to a head to predict a per-frame relevancy score and start/end time displacements. Unlike previous works, our architecture enables Moment Retrieval, Temporal Localization, and Action Segmentation with a single stage model, without the need for action proposals, motion based pretrained features or representation masking. Unlike specialized models, we achieve state of the art results on all three different localization tasks with a unified approach. Code will be available at: \url{https://github.com/google-research/scenic}.

PDF Abstract ICCV 2023 PDF ICCV 2023 Abstract

Code

Add Remove Mark official

google-research/scenic official

↳ Quickstart in

Colab

3,000

Tasks

Add Remove

Action Segmentation

Moment Retrieval

Natural Language Moment Retrieval

Retrieval

Temporal Action Localization

Temporal Localization

Zero-Shot Action Detection

Datasets

ActivityNet

HowTo100M

ActivityNet Captions

Charades-STA COIN

QVHighlights

Results from the Paper

Edit

Ranked #1 on Action Segmentation on COIN

Get a GitHub badge

Task	Dataset	Model	Metric Name	Metric Value	Global Rank	Benchmark
Zero-Shot Action Detection	ActivityNet-1.3	UnLoc-B (75% seen split)	mAP IOU@0.5	40.2	# 3	Compare
Zero-Shot Action Detection	ActivityNet-1.3	UnLoc-B (50% seen split)	mAP IOU@0.5	36.9	# 6	Compare
Zero-Shot Action Detection	ActivityNet-1.3	UnLoc-L (50% seen split)	mAP IOU@0.5	43.2	# 2	Compare
Zero-Shot Action Detection	ActivityNet-1.3	UnLoc-L (75% seen split)	mAP IOU@0.5	47.4	# 1	Compare
Temporal Action Localization	ActivityNet-1.3	UnLoc-L	mAP IOU@0.5	59.3	# 5	Compare
Natural Language Moment Retrieval	ActivityNet Captions	UnLoc-B	R@1,IoU=0.5	48.0	# 4	Compare
			R@1,IoU=0.7	29.7	# 4	Compare
			R@5,IoU=0.5	81.5	# 1	Compare
			R@5,IoU=0.7	61.4	# 2	Compare
Natural Language Moment Retrieval	ActivityNet Captions	UnLoc-L	R@1,IoU=0.5	48.3	# 3	Compare
			R@1,IoU=0.7	30.2	# 2	Compare
			R@5,IoU=0.5	79.2	# 3	Compare
			R@5,IoU=0.7	61.3	# 3	Compare
Moment Retrieval	Charades-STA	UnLoc-L	R@1 IoU=0.5	60.8	# 4	Compare
			R@1 IoU=0.7	38.4	# 5	Compare
			R@5 IoU=0.5	88.2	# 4	Compare
			R@5 IoU=0.7	61.1	# 2	Compare
Moment Retrieval	Charades-STA	UnLoc-B	R@1 IoU=0.5	58.1	# 9	Compare
			R@1 IoU=0.7	35.4	# 10	Compare
			R@5 IoU=0.5	87.4	# 5	Compare
			R@5 IoU=0.7	59.1	# 3	Compare
Action Segmentation	COIN	UnLoc-L	Frame accuracy	72.8	# 1	Compare
Moment Retrieval	QVHighlights	UnLoc-L	R@1 IoU=0.5	66.1	# 4	Compare
Moment Retrieval	QVHighlights	UnLoc-L	R@1 IoU=0.7	46.7	# 12	Compare
Moment Retrieval	QVHighlights	UnLoc-B	R@1 IoU=0.5	64.5	# 8	Compare
Moment Retrieval	QVHighlights	UnLoc-B	R@1 IoU=0.7	48.8	# 5	Compare

Methods

Add Remove

CLIP

Edit Social Preview

UnLoc: A Unified Framework for Video Localization Tasks

Code Edit Add Remove Mark official

Tasks Edit Add Remove

Datasets Edit

Results from the Paper Edit

Methods Edit Add Remove

Code

Add Remove Mark official

Tasks

Add Remove

Datasets

Results from the Paper

Edit

Methods

Add Remove