Iclr 2024 Object Detection. Object detection attacks (szegedy et al., 2014) aim to deceive detectors into making wrong predictions on objects by mainly generating multiple bounding boxes for additional object. Deformable detr can achieve better performance than detr (especially on small objects) with 10$\times$ less training epochs.
Point cloud completion 3d object detectors effectively tackle the challenge of incomplete shapes in sparse point clouds by generating pseudo points to improve detection. Published as a conference paper at iclr 2024 — we propose vwa, a relational representation learner, allowing for varying context window sizes toward multiple receptive.