1: \begin{abstract}

2: DETR has set up a simple end-to-end pipeline for object detection by formulating this task as a set prediction problem, showing promising potential. However, despite the significant progress in improving DETR, this paper identifies a problem of misalignment in the output distribution, which prevents the best-regressed samples from being assigned with  high confidence, hindering the model's accuracy. We propose a metric, recall of best-regressed samples, to quantitively evaluate the misalignment problem. Observing its importance, we  propose a novel \methodname~that incorporates a localization precision aware classification loss in optimization. The proposed loss, IA-BCE, guides the training of DETR to build a strong correlation between classification score and localization precision. We also adopt the mixed-matching strategy, to facilitate DETR-based detectors with faster training convergence while keeping an end-to-end scheme. Moreover, to overcome the dramatic decrease in sample quality induced by the sparsity of queries, we introduce a prime sample weighting mechanism to suppress the interference of unimportant samples. Extensive experiments are conducted with very competitive results reported. In particular, it delivers a $46~(+3.8) \%$~AP on the DAB-DETR baseline with the ResNet-50 backbone and reaches a new SOTA performance of~$50.2\%$ AP in the 1x setting on the COCO validation set when employing the strong baseline DINO. Our code is available at \href{https://github.com/FelixCaae/AlignDETR}{\textcolor{magenta}{https://github.com/FelixCaae/AlignDETR}}.

3: 

4: 

5: 

6: \end{abstract}

7: