Constrained-Focal-Loss Based Deep Learning for Segmentation of Spores

Abstract

The statistics of disease spores is significant for early strategy design of disease control in precision agriculture. To obtain the statistics information of spores in microscopic images, it is crucial to segment spores from images. In this paper, we research a deep learning based method to segment spores, taking anthrax spores as the research objects. We first built an anthrax spore dataset consisting of more than 40,000 spores with accurate labeled spore boundaries to advance the state of the art technology of spore statistics. Then on consideration of the complex class imbalances in actual anthrax spore images, we investigate how class imbalances and hard examples simultaneously influence the loss during training and we discover that hard examples are more likely to appear at the pixels of rare pixels, such as small class pixels and contour pixels. Based on this discovery, we propose Constrained Focal Loss (CFL), which focuses on small class objects, and has a constrained term related to hard examples. In addition, we further propose CFL * , where high importance is put on the pixels surrounding spore contours to improve classification accuracy. The results show that the mean IoU of the DeepLabv3+ trained with CFL * (called as CFL * Net) achieves 91.0%, higher than original DeepLabv3+ with cross-entropy by 8.6 points, and the DeepLabv3+ with Focal Loss by 10.4 points. Moreover, CFLNet * can achieve better performance than original DeepLabv3+, using less than one-third of the training samples and half of the training steps. INDEX TERMS Image segmentation, class imbalance, focal Loss, hard example, convolutional neural networks (CNN).

Dynamically balancing class losses in imbalanced deep learning

“…We propose to dynamically balance class losses (DBCL), which first suppresses the importance of majority class, and then gradually restores the importance of majority class. In the task of classification, CE is widely used, as shown in Equation (5).…”

“…The former increases the learning opportunities of minority classes by over-sampling minority classes or under-sampling majority classes [3]. The latter puts more emphasis on minority classes through cost-sensitive strategies [4][5][6][7]. Owing to intuitiveness and easy implement, the cost-sensitive imbalanced learning methods are widely used.…”

“…Owing to intuitiveness and easy implement, the cost-sensitive imbalanced learning methods are widely used. The existing cost-sensitive methods adopt static balance strategies during training, which emphasizes/suppresses the importance of minority/majority classes from beginning to end in a fixed manner [4][5][6][7][8].…”

“…By adding the idea of DBCL into three state-of-the-art cost-sensitive loss functions [5][6][7], we propose three novel loss functions: dynamic focal-constrained-loss (DCFL), dynamic balance-cross-entropy based on effective number of samples (DBENS) and dynamic balance-crossentropy based on class frequencies (DBCEF), and we adopt them in image salient-instance segmentation. We conduct our experiments on a public dataset, compare DBCL with the state-of-the-art cost-sensitive methods, and evaluate them using mAP (mean average precision) and F1-Score.…”

A Residual Learning-Based Network Intrusion Detection System

“…Focal loss was proposed by He et al [21] in 2017. Focal loss takes the different level of training difficulty of samples into consideration and focuses more on the difficult-to-train samples; therefore, it has been applied in many fields, such as object detection, imbalanced data classification, and so on [22][23][24]. To identify classes with fewer training samples more accurately, a modified focal loss function is used to replace cross-entropy loss function in the proposed model.…”