Abstract
Relying on large scale labeled datasets, deep learning has achieved good performance in image classification tasks. In agricultural and biological engineering, image annotation is time-consuming and expensive. It also requires annotators to have technical skills in specific areas. Obtaining the ground truth is difficult because natural images are expensive. In addition, images in these areas are usually stored as multichannel images, such as computed tomography (CT) images, magnetic resonance images (MRI), and hyperspectral images (HSI). In this paper, we present a framework using active learning and deep learning for multichannel image classification. We use three active learning algorithms, including least confidence, margin sampling, and entropy, as the selection criteria. Based on this framework, we further introduce an “image pool” to make full advantage of images generated by data augmentation. To prove the availability of the proposed framework, we present a case study on agricultural hyperspectral image classification. The results show that the proposed framework achieves better performance compared with the deep learning model. Manual annotation of all the training sets achieves an encouraging accuracy. In comparison, using active learning algorithm of entropy and image pool achieves a similar accuracy with only part of the whole training set manually annotated. In practical application, the proposed framework can remarkably reduce labeling effort during the model development and upadting processes, and can be applied to multichannel image classification in agricultural and biological engineering.
Highlights
Deep convolutional neural networks (CNNs) have achieved outstanding performance in image classification tasks, due to sufficient computing power and well-trained models, and thanks to large scale annotated datasets, such as ImageNet [1], Open Image [2] and PASCALVOC [3]
In order to solve the problem of expensive annotated datasets in agricultural and biological engineering, we present a framework for multichannel images, using active learning algorithm and deep learning framework with an “image pool”
We have introduced deep learning techniques into the classification tasks of agricultural engineering based on the hyperspectral transmittance images, achieving better performance than traditional machine learning methods and proving the feasibility of using CNN to solve multichannel image classification task [28]
Summary
Deep convolutional neural networks (CNNs) have achieved outstanding performance in image classification tasks, due to sufficient computing power and well-trained models, and thanks to large scale annotated datasets, such as ImageNet [1], Open Image [2] and PASCAL. Manual annotation work, which is tedious and time-consuming, can be accomplished by people with limited training. Active learning can achieve better performance with fewer annotated training data since it chooses more informative data to learn [4]. When the unlabeled data are abundant and it is costly to obtain the labels, active learning can possibly build a cost-effective model to significantly reduce the annotation cost. We aim to establish a framework to remarkably reduce the annotation cost without lowering the classification performance, using active learning and deep learning for multichannel image classification
Sign-in/Register to view highlights & summary 
Published Version (
Free)
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call