人工智能
模式识别(心理学)
计算机科学
卷积神经网络
图像(数学)
分割
边距(机器学习)
转化(遗传学)
编码(集合论)
机器学习
生物化学
基因
集合(抽象数据类型)
化学
程序设计语言
作者
Konstantinos Panagiotis Alexandridis,Shan Luo,A. D. Nguyen,Jiankang Deng,Stefanos Zafeiriou
出处
期刊:IEEE transactions on image processing
[Institute of Electrical and Electronics Engineers]
日期:2023-01-01
卷期号:32: 5721-5736
标识
DOI:10.1109/tip.2023.3321461
摘要
The long-tailed distribution is a common phenomenon in the real world. Extracted large scale image datasets inevitably demonstrate the long-tailed property and models trained with imbalanced data can obtain high performance for the over-represented categories, but struggle for the under-represented categories, leading to biased predictions and performance degradation. To address this challenge, we propose a novel de-biasing method named Inverse Image Frequency (IIF) . IIF is a multiplicative margin adjustment transformation of the logits in the classification layer of a convolutional neural network. Our method achieves stronger performance than similar works and it is especially useful for downstream tasks such as long-tailed instance segmentation as it produces fewer false positive detections. Our extensive experiments show that IIF surpasses the state of the art on many long-tailed benchmarks such as ImageNet-LT, CIFAR-LT, Places-LT and LVIS, reaching 55.8% top-1 accuracy with ResNet50 on ImageNet-LT and 26.3% segmentation AP with MaskRCNN ResNet50 on LVIS. Code available at https://github.com/kostas1515/iif.
科研通智能强力驱动
Strongly Powered by AbleSci AI