TY - GEN
T1 - DEEP OBJECT DETECTION WITH EXAMPLE ATTRIBUTE BASED PREDICTION MODULATION
AU - Wu, Zhihao
AU - Liu, Chengliang
AU - Huang, Chao
AU - Wen, Jie
AU - Xu, Yong
N1 - Publisher Copyright:
© 2022 IEEE
PY - 2022
Y1 - 2022
N2 - Deep object detectors suffer from the gradient contribution imbalance during training. In this paper, we point out that such imbalance can be ascribed to the imbalance in example attributes, e.g., difficulty and shape variation degree. We further propose example attribute based prediction modulation (EAPM) to address it. In EAPM, first, the attribute of an example is defined by the prediction and the corresponding ground truth. Then, a modulating factor w.r.t the example attribute is introduced to modulate the prediction error. Finally, the new prediction and the ground-truth are input into the loss function. Essentially, we adjust the gradients of examples with specific attributes to reweight their contribution on the global gradients. We apply EAPM with focal loss and balanced L1 loss to simultaneously solve the imbalance in classification and localization. The experimental results on MS COCO demonstrate that EAPM can bring substantial improvement for deep object detectors.
AB - Deep object detectors suffer from the gradient contribution imbalance during training. In this paper, we point out that such imbalance can be ascribed to the imbalance in example attributes, e.g., difficulty and shape variation degree. We further propose example attribute based prediction modulation (EAPM) to address it. In EAPM, first, the attribute of an example is defined by the prediction and the corresponding ground truth. Then, a modulating factor w.r.t the example attribute is introduced to modulate the prediction error. Finally, the new prediction and the ground-truth are input into the loss function. Essentially, we adjust the gradients of examples with specific attributes to reweight their contribution on the global gradients. We apply EAPM with focal loss and balanced L1 loss to simultaneously solve the imbalance in classification and localization. The experimental results on MS COCO demonstrate that EAPM can bring substantial improvement for deep object detectors.
KW - Deep object detection
KW - classification
KW - example attribute
KW - gradient contribution imbalance
KW - localization
UR - https://www.scopus.com/pages/publications/85131231068
U2 - 10.1109/ICASSP43922.2022.9746194
DO - 10.1109/ICASSP43922.2022.9746194
M3 - 会议稿件
AN - SCOPUS:85131231068
T3 - ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings
SP - 2020
EP - 2024
BT - 2022 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2022 - Proceedings
PB - Institute of Electrical and Electronics Engineers Inc.
T2 - 2022 IEEE International Conference on Acoustics, Speech and Signal Processing, ICASSP 2022
Y2 - 22 May 2022 through 27 May 2022
ER -