LaMI-DETR: Open-Vocabulary Detection with Language Model Instruction

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Du, Penghui, Wang, Yu, Sun, Yifan, Wang, Luting, Liao, Yue, Zhang, Gang, Ding, Errui, Wang, Yan, Wang, Jingdong, Liu, Si
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866917726378786816
author Du, Penghui
Wang, Yu
Sun, Yifan
Wang, Luting
Liao, Yue
Zhang, Gang
Ding, Errui
Wang, Yan
Wang, Jingdong
Liu, Si
author_facet Du, Penghui
Wang, Yu
Sun, Yifan
Wang, Luting
Liao, Yue
Zhang, Gang
Ding, Errui
Wang, Yan
Wang, Jingdong
Liu, Si
contents Existing methods enhance open-vocabulary object detection by leveraging the robust open-vocabulary recognition capabilities of Vision-Language Models (VLMs), such as CLIP.However, two main challenges emerge:(1) A deficiency in concept representation, where the category names in CLIP's text space lack textual and visual knowledge.(2) An overfitting tendency towards base categories, with the open vocabulary knowledge biased towards base categories during the transfer from VLMs to detectors.To address these challenges, we propose the Language Model Instruction (LaMI) strategy, which leverages the relationships between visual concepts and applies them within a simple yet effective DETR-like detector, termed LaMI-DETR.LaMI utilizes GPT to construct visual concepts and employs T5 to investigate visual similarities across categories.These inter-category relationships refine concept representation and avoid overfitting to base categories.Comprehensive experiments validate our approach's superior performance over existing methods in the same rigorous setting without reliance on external training resources.LaMI-DETR achieves a rare box AP of 43.4 on OV-LVIS, surpassing the previous best by 7.8 rare box AP.
format Preprint
id arxiv_https___arxiv_org_abs_2407_11335
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle LaMI-DETR: Open-Vocabulary Detection with Language Model Instruction
Du, Penghui
Wang, Yu
Sun, Yifan
Wang, Luting
Liao, Yue
Zhang, Gang
Ding, Errui
Wang, Yan
Wang, Jingdong
Liu, Si
Computer Vision and Pattern Recognition
Existing methods enhance open-vocabulary object detection by leveraging the robust open-vocabulary recognition capabilities of Vision-Language Models (VLMs), such as CLIP.However, two main challenges emerge:(1) A deficiency in concept representation, where the category names in CLIP's text space lack textual and visual knowledge.(2) An overfitting tendency towards base categories, with the open vocabulary knowledge biased towards base categories during the transfer from VLMs to detectors.To address these challenges, we propose the Language Model Instruction (LaMI) strategy, which leverages the relationships between visual concepts and applies them within a simple yet effective DETR-like detector, termed LaMI-DETR.LaMI utilizes GPT to construct visual concepts and employs T5 to investigate visual similarities across categories.These inter-category relationships refine concept representation and avoid overfitting to base categories.Comprehensive experiments validate our approach's superior performance over existing methods in the same rigorous setting without reliance on external training resources.LaMI-DETR achieves a rare box AP of 43.4 on OV-LVIS, surpassing the previous best by 7.8 rare box AP.
title LaMI-DETR: Open-Vocabulary Detection with Language Model Instruction
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2407.11335