What question did this study set out to answer?

The aim is to develop a more effective method for disrupting transformer-based object detectors using localized query attacks.

March 25, 2026SensorsOpen Access

Localized Query Attack Toward Transformer-Based Visible Object Detectors

View Full Paper

Ask AI

Bookmark

Authors

Innovative attack enhances object detection accuracy in transformer models, suggesting new security measures.

The aim is to develop a more effective method for disrupting transformer-based object detectors using localized query attacks.
Introduced Localized Query Attack (LQA) to focus on specific object features.
Targeted both self-attention in the encoder and cross-attention in the decoder.
Utilized a joint attention matrix to manipulate the influence of encoder outputs.
LQA demonstrated an approximately 20% improvement in transfer attack performance compared to traditional methods.
Real-world validations confirmed the practical effectiveness of LQA.

Wang et al. (2026) studied this question.

View Full Paper

Ask AI

Bookmark

March 25, 2026SensorsOpen Access

View Full Paper

Ask AI

Bookmark

Authors

Innovative attack enhances object detection accuracy in transformer models, suggesting new security measures.

The aim is to develop a more effective method for disrupting transformer-based object detectors using localized query attacks.
Introduced Localized Query Attack (LQA) to focus on specific object features.
Targeted both self-attention in the encoder and cross-attention in the decoder.
Utilized a joint attention matrix to manipulate the influence of encoder outputs.
LQA demonstrated an approximately 20% improvement in transfer attack performance compared to traditional methods.
Real-world validations confirmed the practical effectiveness of LQA.

Wang et al. (2026) studied this question.

View Full Paper

Ask AI

Bookmark