End-to-End Multi-Object Detection with a Regularized Mixture Model

Jaeyoung Yoo, Hojun Lee, Seunghyeon Seo, Inseop Chung, Nojun Kwak
Proceedings of the 40th International Conference on Machine Learning, PMLR 202:40093-40110, 2023.

Abstract

Recent end-to-end multi-object detectors simplify the inference pipeline by removing hand-crafted processes such as non-maximum suppression (NMS). However, during training, they still heavily rely on heuristics and hand-crafted processes which deteriorate the reliability of the predicted confidence score. In this paper, we propose a novel framework to train an end-to-end multi-object detector consisting of only two terms: negative log-likelihood (NLL) and a regularization term. In doing so, the multi-object detection problem is treated as density estimation of the ground truth bounding boxes utilizing a regularized mixture density model. The proposed end-to-end multi-object Detection with a Regularized Mixture Model (D-RMM) is trained by minimizing the NLL with the proposed regularization term, maximum component maximization (MCM) loss, preventing duplicate predictions. Our method reduces the heuristics of the training process and improves the reliability of the predicted confidence score. Moreover, our D-RMM outperforms the previous end-to-end detectors on MS COCO dataset. Code is available at https://github.com/lhj815/D-RMM.

Cite this Paper

BibTeX
@InProceedings{pmlr-v202-yoo23b, title = {End-to-End Multi-Object Detection with a Regularized Mixture Model}, author = {Yoo, Jaeyoung and Lee, Hojun and Seo, Seunghyeon and Chung, Inseop and Kwak, Nojun}, booktitle = {Proceedings of the 40th International Conference on Machine Learning}, pages = {40093--40110}, year = {2023}, editor = {Krause, Andreas and Brunskill, Emma and Cho, Kyunghyun and Engelhardt, Barbara and Sabato, Sivan and Scarlett, Jonathan}, volume = {202}, series = {Proceedings of Machine Learning Research}, month = {23--29 Jul}, publisher = {PMLR}, pdf = {https://proceedings.mlr.press/v202/yoo23b/yoo23b.pdf}, url = {https://proceedings.mlr.press/v202/yoo23b.html}, abstract = {Recent end-to-end multi-object detectors simplify the inference pipeline by removing hand-crafted processes such as non-maximum suppression (NMS). However, during training, they still heavily rely on heuristics and hand-crafted processes which deteriorate the reliability of the predicted confidence score. In this paper, we propose a novel framework to train an end-to-end multi-object detector consisting of only two terms: negative log-likelihood (NLL) and a regularization term. In doing so, the multi-object detection problem is treated as density estimation of the ground truth bounding boxes utilizing a regularized mixture density model. The proposed end-to-end multi-object Detection with a Regularized Mixture Model (D-RMM) is trained by minimizing the NLL with the proposed regularization term, maximum component maximization (MCM) loss, preventing duplicate predictions. Our method reduces the heuristics of the training process and improves the reliability of the predicted confidence score. Moreover, our D-RMM outperforms the previous end-to-end detectors on MS COCO dataset. Code is available at https://github.com/lhj815/D-RMM.} }
Endnote
%0 Conference Paper %T End-to-End Multi-Object Detection with a Regularized Mixture Model %A Jaeyoung Yoo %A Hojun Lee %A Seunghyeon Seo %A Inseop Chung %A Nojun Kwak %B Proceedings of the 40th International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2023 %E Andreas Krause %E Emma Brunskill %E Kyunghyun Cho %E Barbara Engelhardt %E Sivan Sabato %E Jonathan Scarlett %F pmlr-v202-yoo23b %I PMLR %P 40093--40110 %U https://proceedings.mlr.press/v202/yoo23b.html %V 202 %X Recent end-to-end multi-object detectors simplify the inference pipeline by removing hand-crafted processes such as non-maximum suppression (NMS). However, during training, they still heavily rely on heuristics and hand-crafted processes which deteriorate the reliability of the predicted confidence score. In this paper, we propose a novel framework to train an end-to-end multi-object detector consisting of only two terms: negative log-likelihood (NLL) and a regularization term. In doing so, the multi-object detection problem is treated as density estimation of the ground truth bounding boxes utilizing a regularized mixture density model. The proposed end-to-end multi-object Detection with a Regularized Mixture Model (D-RMM) is trained by minimizing the NLL with the proposed regularization term, maximum component maximization (MCM) loss, preventing duplicate predictions. Our method reduces the heuristics of the training process and improves the reliability of the predicted confidence score. Moreover, our D-RMM outperforms the previous end-to-end detectors on MS COCO dataset. Code is available at https://github.com/lhj815/D-RMM.
APA
Yoo, J., Lee, H., Seo, S., Chung, I. & Kwak, N.. (2023). End-to-End Multi-Object Detection with a Regularized Mixture Model. Proceedings of the 40th International Conference on Machine Learning, in Proceedings of Machine Learning Research 202:40093-40110 Available from https://proceedings.mlr.press/v202/yoo23b.html.

Related Material