MM-RCNN: Toward Few-Shot Object Detection in Remote Sensing Images With Meta Memory

Li, Jianxiang; Tian, Yan; Xu, Yiping; Hu, Xinli; Zhang, Zili; Wang, Hu; Xiao, Yiwen

doi:10.1109/tgrs.2022.3228612

Cited by 20 publications

(5 citation statements)

References 56 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The second is FeatReweighting [48], which uses feature weighting to achieve FSOD; The third method uses fine-tuning for the TFA [51] to improve the accuracy of few-shot object detection; The fourth Meta R-CNN [49] adopts a meta-learning approach to enhance the recognition of novel classes by using class attention vectors; The fifth FSCE [52] introduces the method of contrastive learning embedding to realize FSOD. Both P-CNN [71] and MM-RCNN [55] are few-shot detection methods for remote sensing images.…”

Section: B Performance Analysis On Dior Datasetmentioning

confidence: 99%

“…Recent studies have shown a growing interest in the field of few-shot remote sensing image object detection [53]- [55]. In traditional few-shot object detection approaches for Remote Sensing Imagery (RSI), the conversion of support images from class-agnostic features to class-specific vectors followed by feature attention operations on query image features has been the norm.…”

Section: Introductionmentioning

confidence: 99%

“…In traditional few-shot object detection approaches for Remote Sensing Imagery (RSI), the conversion of support images from class-agnostic features to class-specific vectors followed by feature attention operations on query image features has been the norm. However, these methods encounter two significant challenges: They often overlook the spatial similarity between support and query features, which is crucial for effective RSI detection [53]- [55]. Feature attention operations are typically unidirectional, resulting in asymmetric learned support-query relations.…”

Section: Introductionmentioning

confidence: 99%

“…This method combines unbiased intersection over union (IoU) scores and foregroundbackground confidence scores to filter out redundant region proposals while retaining more foreground proposals for novel categories in the RPN. Lu et al [55] have proposed a metalearning-based approach named Meta-Memory-based RCNN (MM-RCNN). This approach utilizes a memory module to store knowledge acquired for each category during training and employs a memory-based external attention (MEA) mech-anism to aggregate category information simultaneously.…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Few-Shot Object Detection Based on Contrastive Class-Attention Feature Reweighting for Remote Sensing Images

Miao,

Zhao,

Geng

et al. 2024

IEEE J. Sel. Top. Appl. Earth Observations Remote Sensing

View full text Add to dashboard Cite

Remote sensing image object detection with deep neural networks has been highly successful, but it heavily relies on a large number of labeled samples for optimal performance. Unfortunately, when faced with limited labeled samples, the performance of object detection deteriorates. In order to overcome these limitations, we propose a few-shot object detection (FSOD) method based on the reweighting of contrastive class-attention features for remote sensing images. A Siamese representation embedding model based on contrastive learning with a distinguishing operator is proposed to deal with inter-class feature ambiguity between base classes and novel classes under complex backgrounds. At the same time, the attention class weights method on Region of Interest (ROI) features is utilized to boost the discrimination of the novel class features. We conducted comprehensive experiments on two widely used remote sensing object detection datasets, RSOD and DIOR. The proposed FSOD model demonstrated superior performance compared to MM-RCNN, achieving an improvement of approximately 4.7% in terms of mAP on the DIOR dataset. Additionally, in comparison to SAAN, the FSOD model exhibited an improvement of approximately 3.4% in mAP on the RSOD dataset.

show abstract

Section: B Performance Analysis On Dior Datasetmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Few-Shot Object Detection Based on Contrastive Class-Attention Feature Reweighting for Remote Sensing Images

Miao,

Zhao,

Geng

et al. 2024

IEEE J. Sel. Top. Appl. Earth Observations Remote Sensing

View full text Add to dashboard Cite

show abstract

“…In the last few years, deep learning has achieved a good performance in satellite image processing tasks, including image fusion, image classification, object detection, and picture reconstruction [14][15][16][17][18][19][20][21][22][23]. It employs high-parameter models to comprehend complex textures and spectral features for image interpolation tasks, using the images from two time phases to output the intermediate phase image.…”

Section: Introductionmentioning

confidence: 99%

Enhancing Satellite Image Sequences through Multi-Scale Optical Flow-Intermediate Feature Joint Network

Shi,

Liu,

Zhang

et al. 2024

Remote Sensing

View full text Add to dashboard Cite

Satellite time-series data contain information in three dimensions—spatial, spectral, and temporal—and are widely used for monitoring, simulating, and evaluating Earth activities. However, some time-phase images in the satellite time series data are missing due to satellite sensor malfunction or adverse atmospheric conditions, which prevents the effective use of the data. Therefore, we need to complement the satellite time series data with sequence image interpolation. Linear interpolation methods and deep learning methods that have been applied to sequence image interpolation lead to large errors between the interpolation results and the real images due to the lack of accurate estimation of pixel positions and the capture of changes in objects. Inspired by video frame interpolation, we combine optical flow estimation and deep learning and propose a method named Multi-Scale Optical Flow-Intermediate Feature Joint Network. This method learns pixel occlusion and detailed compensation information for each channel and jointly refines optical flow and intermediate features at different scales through an end-to-end network together. In addition, we set a spectral loss function to optimize the network’s learning of the spectral features of satellite images. We have created a time-series dataset using Landsat-8 satellite data and Sentinel-2 satellite data and then conducted experiments on this dataset. Through visual and quantitative evaluation of the experimental results, we discovered that the interpolation results of our method retain better spectral and spatial consistency with the real images, and that the results of our method on the test dataset have a 7.54% lower Root Mean Square Error than other approaches.

show abstract

DBFGAN: Dual Branch Feature Guided Aggregation Network for remote sensing image

Chu

Xia

et al. 2023

International Journal of Applied Earth Observation and Geoinfor

View full text Add to dashboard Cite

MM-RCNN: Toward Few-Shot Object Detection in Remote Sensing Images With Meta Memory

Cited by 20 publications

References 56 publications

Few-Shot Object Detection Based on Contrastive Class-Attention Feature Reweighting for Remote Sensing Images

Few-Shot Object Detection Based on Contrastive Class-Attention Feature Reweighting for Remote Sensing Images

Enhancing Satellite Image Sequences through Multi-Scale Optical Flow-Intermediate Feature Joint Network

DBFGAN: Dual Branch Feature Guided Aggregation Network for remote sensing image

Contact Info

Product

Resources

About