Object Detection in High-Resolution Remote Sensing Images Based on a Hard-Example-Mining Network

Zhang, Lei; Wang, Yuehuan; Huo, Yang

doi:10.1109/tgrs.2020.3038673

Cited by 15 publications

(7 citation statements)

References 43 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Various target detection algorithms are compared with the MashFormer on the HRRSD dataset, including Faster R-CNN [8], Mask R-CNN [9], FCOS [15], MSHEMN [46], SGFTHR [47], and GLFPN [48].…”

Section: Resultsmentioning

confidence: 99%

MashFormer: A Novel Multiscale Aware Hybrid Detector for Remote Sensing Object Detection

Wang

Bai

et al. 2023

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

View full text Add to dashboard Cite

Object detection is a critical and demanding topic in the subject of processing satellite and airborne images. The targets acquired in remote sensing imagery are at various sizes, and the backgrounds are complicated, which makes object detection extremely challenging. We address these aforementioned issues in this paper by introducing the MashFormer, an innovative multi-scale aware CNN and Transformer integrated hybrid detector. Specifically, MashFormer employs the transformer block to complement the convolutional neural network (CNN) based feature extraction backbone, which could obtain the relationships between long-range features and enhance the representative ability in complex background scenarios. With the intention of improving the detection performance for objects with multi-scale characteristic, since in remote sensing scenarios, the size of object varies greatly. A multi-level feature aggregation component, incoperate with a cross-level feature alignment module is designed to alleviate the semantic discrepancy between features from shallow and deep layers. To verify the effectiveness of the suggested MashFormer, comparative experiments are carried out with other cutting-edge methodologies using the publicly available High Resolution Remote Sensing Detection (HRRSD) and Northwestern Polytechnical University (NWPU) VHR-10 datasets. The experimental findings confirm the effectiveness and superiority of our suggested model by indicating that our approach has greater mean average precision (mAP) than the other methodologies.

show abstract

“…Various target detection algorithms are compared with the MashFormer on the HRRSD dataset, including Faster R-CNN [8], Mask R-CNN [9], FCOS [15], MSHEMN [46], SGFTHR [47], and GLFPN [48].…”

Section: Resultsmentioning

confidence: 99%

MashFormer: A Novel Multiscale Aware Hybrid Detector for Remote Sensing Object Detection

Wang

Bai

et al. 2023

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

View full text Add to dashboard Cite

show abstract

“…Although it is not the best in other categories, it still has strong competitiveness. The proposed method is 3.8%, 2.9%, 5.1%, 2.3%, 1.8%, 1.3% higher than RetinaNet, [ 37 ] Faster R‐CNN, [ 22 ] GACL Faster R‐CNN, [ 45 ] Cascade R‐CNN, [ 46 ] Mask R‐CNN, [ 25 ] and MSHEMN, [ 47 ] respectively. Except for GACL Faster R‐CNN, the former has obvious speed advantages.…”

Section: Results and Analysismentioning

confidence: 99%

Toward Structural Learning and Enhanced YOLOv4 Network for Object Detection in Optical Remote Sensing Images

Wang

Liu

2022

Advcd Theory and Sims

View full text Add to dashboard Cite

With the maturity of technological tools such as satellites and airplanes, object detection in optical remote sensing images have been widely used in military and civilian fields. Due to the interference of extensive multi-scale targets and complex backgrounds, the existing algorithms are still limited, especially for small-scale targets. To solve this problem, a high-precision remote sensing detection method based on the advanced YOLOv4 framework is proposed. First, a clustering algorithm that combines knowledge of object scales to generate a priori anchor boxes with a higher matching degree is proposed. The feature extension module is designed to expand the receptive field of the backbone network and capture important contextual information. After that, coordinate attention (CA) is introduced to suppress the extra noise under multi-scale fusion that is rarely noticed in previous work. In addition, a convolutional idea of self-learning structure has also been novelly proposed and combined with a high-resolution network to efficiently enhance the performance of the network under limited computing resources. Finally, detailed experiments are performed on the HRRSD dataset. The experimental results show that the network has a more satisfactory superiority, and it is conducive to the advancement of optical remote sensing technology.

show abstract

“…Then, in the second step, HEG mined a fixed number of hard examples from amounts of real and artificially generated examples. In [43], hard example mining was introduced to object detection in HSR RSIs in a cascaded manner to make the network focus on the hard examples during training by changing the distribution of input data for each stage. In a detection problem, hard examples correspond to false positive detection.…”

Section: Online Hard Example Miningmentioning

confidence: 99%

$\text{A}^{3}$Seg: An Annealing Augmented and Aligned Semantic Segmentation Network for High Spatial Resolution Remote Sensing Images

Xue

Liu

Huang

et al. 2022

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

View full text Add to dashboard Cite

Semantic segmentation of high spatial resolution (HSR) remote sensing images (RSIs) plays an important role in many applications. However, HSR RSIs have significantly larger spatial sizes than typical natural images, which results in fewer valuable samples when training models. In addition, fusing multiscale features is the key step in obtaining features with strong semantic and high spatial information. However, current feature fusion methods are too straightforward to address misalignment issues. To handle these two problems, we propose an annealing augmented and aligned segmentation network, named A 3 Seg. Specifically, we propose an annealing online hard example mining (AOHEM) strategy to automatically select more valuable samples during the training stage. Based on AOHEM, a contextual augmentation block is proposed to extract sufficient contextual information using three different attention mechanisms in consideration of three different feature properties. Finally, we propose a novel feature alignment block to fuse features at different levels by alignment with the guidance of a salient feature. Experimental results on three different HSR RSIs datasets demonstrate that the proposed method outperforms the state-of-the-art general semantic segmentation methods with a better tradeoff between accuracy and complexity.

show abstract

Object Detection in High-Resolution Remote Sensing Images Based on a Hard-Example-Mining Network

Cited by 15 publications

References 43 publications

MashFormer: A Novel Multiscale Aware Hybrid Detector for Remote Sensing Object Detection

MashFormer: A Novel Multiscale Aware Hybrid Detector for Remote Sensing Object Detection

Toward Structural Learning and Enhanced YOLOv4 Network for Object Detection in Optical Remote Sensing Images

$\text{A}^{3}$Seg: An Annealing Augmented and Aligned Semantic Segmentation Network for High Spatial Resolution Remote Sensing Images

Contact Info

Product

Resources

About