Lightweight Tensorized Neural Networks for Hyperspectral Image Classification

Ma, Tian-Yu; Li, Heng-Chao; Wang, Rui; Du, Qian; Jia, Xiuping

doi:10.1109/tgrs.2022.3225438

Cited by 11 publications

(2 citation statements)

References 69 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We use a shrunk PB to constrain the sampling position, making the sampling point better located inside the rotated bounding box. Given an F × W × H (F denotes feature channels) feature cubes X LI , for each position P(x 0 , y 0 ), we obtain ERF transformation results FE LI by FE LI (P) = Deformable(X LI , offsets(P)), (4) where offsets(P) is the position bias with the size of W × H × 18. The original offsets of 3 × 3 convolutions can be defined as O G = {(−1, -1), (−1, 0), (−1, 1), (0, −1), .…”

Section: Erf Transformationmentioning

confidence: 99%

“…As a fundamental task in remote sensing image understanding, arbitrary-orientedobject-detection (AOOD) is attracting the attention of researchers more and more. At the same time, with the rapid development of convolutional neural network (CNN)-based methods [1][2][3][4][5], many outstanding AOOD detectors stand out [6][7][8][9][10][11][12]. However, different from object detection in natural images, AOOD is more challenging mainly due to the following two reasons:…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

AFRE-Net: Adaptive Feature Representation Enhancement for Arbitrary Oriented Object Detection

Zhang,

Sun,

Zhuang

et al. 2023

Remote Sensing

View full text Add to dashboard Cite

Arbitrary-oriented object detection (AOOD) is a crucial task in aerial image analysis but is also faced with significant challenges. In current AOOD detectors, commonly used multi-scale feature fusion modules fall short in spatial and semantic information complement between scales. Additionally, fixed feature extraction structures are usually used following a fusion model, resulting in the inability of detectors to self-adjust. At the same time, feature fusion and extraction modules are designed in isolation and the internal synergy between them is ignored. The above problems result in feature representation deficiency, thus affecting the overall detection precision. To solve these problems, we first create a fine-grained feature pyramid network (FG-FPN) that not only provides richer spatial and semantic features, but also completes neighbor scale features in a self-learning mode. Subsequently, we propose a novel feature enhancement module (FEM) to fit FG-FPN. FEM authorizes the detection unit to automatically adjust the sensing area and adaptively suppress background interference, thereby generating stronger feature representations. Our proposed solution was tested through extensive experiments on challenging datasets, including DOTA (77.44% mAP), HRSC2016 (97.82% mAP), UCAS-AOD (91.34% mAP), as well as ICDAR2015 (86.27% F-score) and its effectiveness and high applicability are verified on all the above datasets.

show abstract

Section: Erf Transformationmentioning

confidence: 99%