Object detection of aerial image using mask-region convolutional neural network (mask R-CNN)

Musyarofah, None; Schmidt, Valentina; Kada, Martin

doi:10.1088/1755-1315/500/1/012090

Cited by 8 publications

(4 citation statements)

References 15 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…• Localization based on rotated bounding boxes often lacks precision due to angle issues. Some works [41], [54], [55] have demonstrated the effectiveness of generating bounding boxes by predicting masks. We will conduct further research in this direction.…”

Section: Discussionmentioning

confidence: 99%

A2Net: An Anchor-Free Alignment Network for Oriented Object Detection in Remote Sensing Images

Yang,

Cao,

Huang

et al. 2024

IEEE Access

View full text Add to dashboard Cite

Object detection in remote sensing images is crucial for identifying and locating objects in the field, holding significance in remote sensing. Oriented object detection employs oriented bounding boxes to locate objects with varying orientations, achieving recent advancements. However, challenges persist due to vast variations in object scale and orientation. Existing methods use Intersection over Union (IoU) to measure bounding box quality but often ignore shape information. Unlike horizontal detectors, oriented detectors always incorporate an angle parameter. Yet, objects with different shapes exhibit varying angle sensitivity. For objects with the same angle but different shapes, their IoU can differ significantly. We argue that relying solely on IoU is not comprehensive. To address this, we propose the Shape-aware IoU Score (SaIS), considering shape information and IoU for each bounding box. We use SaIS to enhance the dynamic soft label assignment strategy, resulting in an improved Shape-aware Label Assignment (SaLA). SaLA aids the detector in selecting more suitable samples. Leveraging RTMDet-R and S2ANet strengths, we design an Anchor-free Alignment Network (A2Net) for oriented object detection. A2Net features two detection heads: the initial head and the refinement head. Utilizing alignment convolution (AlignConv) between these heads obtains aligned features. We validate the proposed approach's effectiveness on the DOTA dataset and DIOR-R dataset.INDEX TERMS Oriented object detection, shape-aware, remote sensing, deep learning.

show abstract

Section: Discussionmentioning

confidence: 99%

A2Net: An Anchor-Free Alignment Network for Oriented Object Detection in Remote Sensing Images

Yang,

Cao,

Huang

et al. 2024

IEEE Access

View full text Add to dashboard Cite

show abstract

“…Its construction utilises a fully connected convolutional object detection network with an added branch, identifying and predicting segmentation masks on regions of interest defined by the object detection network. This more widely connected convolutional network is then passed, via a backbone, to a deep neural network for instance classification (Musyarofah et al, 2020;Wang et al, 2020).…”

Section: Literature Review and Related Studiesmentioning

confidence: 99%

The Use of Computer Vision to Combat Losses from Disease in Grapevines

2023

CEIS

View full text Add to dashboard Cite

The use of computer vision to support and automate agriculture and viticulture is increasing. Therefore, it is important to continuously test new technologies and equipment. Management of pests and diseases in viticulture is a labour-intensive task. This study aims to investigate current technologies in computer vision that could be applied to disease and pest detection in viticulture and the application of transfer learning on segmentation networks. This study also implements a case study and applies computer vision for disease and pest detection. Observation of limitations in the network's performance on testing images, after training on the limited data set, suggests that careful control is needed over lighting conditions in the image capture environment. Although initial results are positive, a larger training dataset is recommended to achieve a greater level of accuracy.

show abstract

“…DL-based techniques can solve several problems in Geosciences. Among those problems we can cite object detection [139,140], hyperspectral image classification [10,141], super-resolution [142,143,144], change detection [145,146] and semantic segmentation.…”

Section: Applications On Remote Sensing and Examples Of Available Datasetsmentioning

confidence: 99%

A Review of Remote Sensing Applications on Very High-Resolution Imagery Using Deep Learning-Based Semantic Segmentation Techniques

Borba¹,

Bias²,

Silva³

et al. 2021

IJAERS

View full text Add to dashboard Cite

Semantic Segmentation is a technique in Computer Sciences(CS) to extract information from images. Recent advances in Artificial Intelligence, particularly in Deep Learning, Semantic Segmentation combined with techniques such as convolutional neural networks, have presented better results and exciting results. Due to its power and better results than classical approaches, there has been an increase in research articles in Remote Sensing that propose using deep learning-based semantic Segmentation to extract information from satellite or airborne imagery. In this paper, we surveyed the state-of-the-art of Semantic Segmentation in Remote Sensing from 2010 until 2020 by identifying the research topics and the number of publications and citations. Furthermore, we also pointed out the fundamental algorithms, the main convolutional neural network architectures, backbones, and the most used evaluation metrics. In addition, some datasets were highlighted, as well as some frameworks that can be used to train semantic segmentation deep neural networks. Finally, we have shown some applications of the showcased techniques and concluded the paper by pointing out some research opportunities of Remote Sensing Semantic Segmentation, concerning some bleeding-edge scientific papers published in 2020 in CS.

show abstract

Object detection of aerial image using mask-region convolutional neural network (mask R-CNN)

Cited by 8 publications

References 15 publications

A2Net: An Anchor-Free Alignment Network for Oriented Object Detection in Remote Sensing Images

A2Net: An Anchor-Free Alignment Network for Oriented Object Detection in Remote Sensing Images

The Use of Computer Vision to Combat Losses from Disease in Grapevines

A Review of Remote Sensing Applications on Very High-Resolution Imagery Using Deep Learning-Based Semantic Segmentation Techniques

Contact Info

Product

Resources

About