Small Object Detection using Context and Attention 论文阅读笔记 出处:2021 International Conference on Artificial Intelligence in Information and Communication (ICAIIC) Jeju Island, Korea 一、介绍 目标检测算法在各种环境下的应用存在许多局限性,特别是检测小物体仍然具有挑战性,因为它们的分辨率低,信息有限。 文章...
1 Introduction 首先给小目标提供足够多的信息,通过从高层中提取背景信息给小目标,然后拼接特征层,就能增强小目标的信息。然后为了让网络聚焦于小目标,在浅层增加了注意力机制,能减少网络对于背景的关注。 2 …
Image color analysis,Object detection,Detectors,Artificial intelligence,Context modelingThere are many limitations applying object detection algorithm on various environments. Specifically, detecting small objects is still challenging because they have low-resolution and limited information. We propose an object...
Small object detectionDual-branch feature extraction networkBoundary context-awareness moduleSelf-attention frequency-refinement moduleDue to the high flight altitude and large reconnaissance area of Unmanned Aerial Vehicle, objects in aerial images usually have limited feature information and low resolution, ...
HCF-Net: Hierarchical Context Fusion Network for Infrared Small Object Detection 解释HCF-Net是什么: HCF-Net(Hierarchical Context Fusion Network)是一种专为红外小目标检测设计的深度学习模型。它旨在通过融合不同层次的上下文信息来提高红外图像中小目标的检测性能。HCF-Net通过构建一个多层次的特征提取和融合框...
Infrared small object detection is an important computer vision task involving the recognition and localization of tiny objects in infrared images, which usually contain only a few pixels. However, it encounters difficulties due to the diminutive size of the objects and the generally complex backgrounds...
On the newly released COCO dataset, our models provide relative improvements of up to 5% over CNN-based state-of-the-art detectors, with the gains concentrated on hard cases such as small objects (10% relative improvement).doi:10.1371/journal.pone.0098447Gupta, Saurabh...
Especially, the Average Precision (AP) metric is significantly improved by 17.0% on small object detection on MSCOCO. Introduction For humans, it is easy to precisely locate and classify the objects of interest of an image in real time. Correspondingly, many real-world object detection scenarios...
我们的Container架构在ImageNet上使用22M参数实现了82.7%的Top-1精度,比DeiT-Small提高了2.8,并且可以在短短200个时代收敛到79.9%的Top-1精度。比起相比的基于Transformer的方法不能很好地扩展到下游任务依赖较大的输入图像的分辨率,我们高效的网络,名叫CONTAINER-LIGHT,可以使用在目标检测和分割网络如DETR实例,RetinaNet...
这种操作类似relu操作。 总结: (1), paper 使用了multi-scale 进行object detection,在浅层Conv层对其feature maps进行roi-pooling, 增强了对small object的detect能力。 (2),使用了RNN对其周围的region的信息,增强feature信息,促进classification。