Few-Shot Object Detection on Remote Sensing Images via Shared Attention Module and Balanced Fine-Tuning Strategy

Huang, Xu; He, Bokun; Tang, Ming; Wang, Dingwen; He, Chu

doi:10.3390/rs13193816

Cited by 26 publications

(41 citation statements)

References 27 publications

(19 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Next, k annotated samples are selected to constitute the support set for k-shot learning. Because k is typically used in the range of 0-20 in the dataset [70], [71], we chose 0, 5, 10, and 20 for samples annotated with k. Lastly, we have trained zeroshot, 5-shot, 10-shot, and 20-shot in trained classifiers [40], [72]. The support set was used to optimize the classifier for 60 iterations, where the batch size was 1.…”

Section: Few Shot Learning Using Independent Databasementioning

confidence: 99%

Automatic Cardiac Arrhythmia Classification Using Residual Network Combined With Long Short-Term Memory

Kim

Lee

Song³

et al. 2022

IEEE Trans. Instrum. Meas.

View full text Add to dashboard Cite

Diagnosis and classification of arrhythmia, which is associated with abnormal electrical activities in the heart, are critical for clinical treatments. Previous studies focused on the diagnosis of atrial fibrillation, which is the most common arrhythmia in adults. The classification performance achieved by studies on other arrhythmia types is not satisfactory for clinical use owing to the small number of classes (minority classes). In this study, we propose a novel framework for automatic classification that combines a residual network with a squeeze and excitation block, and a bidirectional long short-term memory. 8-class, 4-class, and 2-class performances were evaluated on the MIT-BIH arrhythmia database (MITDB), MIT-BIH atrial fibrillation database (AFDB), and PhysioNet/Computing in the cardiology challenge 2017 database (CinC DB), respectively, and they were superior to performance achieved by conventional methods. In addition, the class-wise F1-score in the minority classes was higher than those of the methods adopted in existing studies. To measure the generalization ability of the proposed framework, AFDB and CinC DB were tested using a MITDBtrained model, and superior performance was achieved compared with ShallowConvNet and DeepConvNet. We performed a crosssubject experiment using AFDB and obtained a statistically higher performance using the proposed method compared with typical machine learning methods. The proposed framework can enable the direct diagnosis of arrhythmia types in clinical trials based on the accurate detection of the minority class.

show abstract

Section: Few Shot Learning Using Independent Databasementioning

confidence: 99%

Automatic Cardiac Arrhythmia Classification Using Residual Network Combined With Long Short-Term Memory

Kim

Lee

Song³

et al. 2022

IEEE Trans. Instrum. Meas.

View full text Add to dashboard Cite

show abstract

“…To allow the deep model to generalize in the target domain with only a few samples, the researchers propose a new machine learning method, namely few-shot learning (FSL) [ 33 , 34 , 35 , 36 , 37 ] for this problem. However, insufficient samples will bring difficulties in model training, resulting in overfitting.…”

Section: Related Workmentioning

confidence: 99%

MSFFAL: Few-Shot Object Detection via Multi-Scale Feature Fusion and Attentive Learning

Zhang

Sun

Wan

et al. 2023

Sensors

View full text Add to dashboard Cite

Few-shot object detection (FSOD) is proposed to solve the application problem of traditional detectors in scenarios lacking training samples. The meta-learning methods have attracted the researchers’ attention for their excellent generalization performance. They usually select the same class of support features according to the query labels to weight the query features. However, the model cannot possess the ability of active identification only by using the same category support features, and feature selection causes difficulties in the testing process without labels. The single-scale feature of the model also leads to poor performance in small object detection. In addition, the hard samples in the support branch impact the backbone’s representation of the support features, thus impacting the feature weighting process. To overcome these problems, we propose a multi-scale feature fusion and attentive learning (MSFFAL) framework for few-shot object detection. We first design the backbone with multi-scale feature fusion and channel attention mechanism to improve the model’s detection accuracy on small objects and the representation of hard support samples. Based on this, we propose an attention loss to replace the feature weighting module. The loss allows the model to consistently represent the objects of the same category in the two branches and realizes the active recognition of the model. The model no longer depends on query labels to select features when testing, optimizing the model testing process. The experiments show that MSFFAL outperforms the state-of-the-art (SOTA) by 0.7–7.8% on the Pascal VOC and exhibits 1.61 times the result of the baseline model in MS COCO’s small objects detection.

show abstract

“…Models are welldesigned network architectures to achieve FSOD tasks, in which different components have different roles, aiming to implement cross-domain mapping of shared semantics. Strategies refer to learning strategies, which means how to conduct a network to implement FSOD, such as meta learning [10,11], transfer learning [16,17], etc. In this paper, we will follow the above clue to review and discuss the research status of few-shot object detection in remote sensing images and look forward to the development prospects.…”

Section: N Base I=1mentioning

confidence: 99%

Few-Shot Object Detection in Remote Sensing Image Interpretation: Opportunities and Challenges

Liu

You

et al. 2022

Remote Sensing

View full text Add to dashboard Cite

Recent years have witnessed rapid development and remarkable achievements on deep learning object detection in remote sensing (RS) images. The growing improvement of the accuracy is inseparable from the increasingly complex deep convolutional neural network and the huge amount of sample data. However, the under-fitting neural network will damage the detection performance facing the difficulty of sample acquisition. Thus, it evolves into few-shot object detection (FSOD). In this article, we first briefly introduce the object detection task and its algorithms, to better understand the basic detection frameworks followed by FSOD. Then, FSOD design methods in RS images for three important aspects, such as sample, model, and learning strategy, are respectively discussed. In addition, some valuable research results of FSOD in computer vision field are also included. We advocate a wide research technique route, and some advice about feature enhancement and multi-modal fusion, semantics extraction and cross-domain mapping, fine-tune and meta-learning strategies, and so on, are provided. Based on our stated research route, a novel few-shot detector that focuses on contextual information is proposed. At the end of the paper, we summarize accuracy performance on experimental datasets to illustrate the achievements and shortcomings of the stated algorithms, and highlight the future opportunities and challenges of FSOD in RS image interpretation, in the hope of providing insights into future research.

show abstract

Few-Shot Object Detection on Remote Sensing Images via Shared Attention Module and Balanced Fine-Tuning Strategy

Cited by 26 publications

References 27 publications

Automatic Cardiac Arrhythmia Classification Using Residual Network Combined With Long Short-Term Memory

Automatic Cardiac Arrhythmia Classification Using Residual Network Combined With Long Short-Term Memory

MSFFAL: Few-Shot Object Detection via Multi-Scale Feature Fusion and Attentive Learning

Few-Shot Object Detection in Remote Sensing Image Interpretation: Opportunities and Challenges

Contact Info

Product

Resources

About