


Remote sensing imaging technology has made significant progress in the past few decades. With the continuous improvement of space, spectrum and resolution of modern airborne sensors, they can cover most of the earth's surface. Therefore, remote sensing technology has many applications in ecology, environmental science, soil science, water pollution, glaciology, land measurement and analysis, etc. The field of research plays a vital role. Because remote sensing data are often multi-modal, located in geographic space (geolocation), often at a global scale, and the data size is growing, these characteristics bring unique challenges to the automatic analysis of remote sensing imaging.
In many areas of computer vision, such as object recognition, detection, segmentation, etc., deep learning, especially convolutional neural networks (CNN), has become mainstream. Convolutional neural networks typically take RGB images as input and perform a series of convolution, local normalization, and pooling operations. CNNs typically rely on large amounts of training data and then use the resulting pre-trained model as a universal feature extractor for various downstream applications. The success of computer vision technology based on deep learning has also inspired the remote sensing community and has made significant progress in many remote sensing tasks, such as hyperspectral image classification and change detection.
One of the main fundamentals of CNN is the convolution operation, which captures local interactions between elements in the input image, such as contour and edge information. CNNs encode biases such as spatial connectivity and translational equivalence, features that help build versatile and efficient architectures. The local receptive field in CNN limits the modeling of long-range dependencies in images (such as relationships between distant parts). Convolution is content-independent because the weights of the convolutional filters are fixed, applying the same weight to all inputs regardless of their nature. Visual transformers (ViTs) have demonstrated impressive performance in a variety of tasks in computer vision. Based on the self-attention mechanism, ViT effectively captures global interactions by learning the relationships between sequence elements. Recent studies have shown that ViT has content-dependent long-range interaction modeling capabilities and can flexibly adjust its receptive fields to combat interference in data and learn effective feature representations. As a result, ViT and its variants have been successfully used in many computer vision tasks, including classification, detection, and segmentation.
With the success of ViT in the field of computer vision, the number of tasks using the transformer framework based on remote sensing analysis has increased significantly (see Figure 1), such as ultra-high-resolution image classification, change detection, Transformers are used for full-color sharpening, building detection and image subtitles. This opens a new era of remote sensing analysis, with researchers using various methods such as leveraging ImageNet pre-training or using visual transformers to perform remote sensing pre-training.
#Similarly, there are approaches in the literature based on pure transformer designs or utilizing hybrid approaches based on transformers and CNNs. Due to the rapid emergence of transformer-based methods for different remote sensing problems, it is becoming increasingly challenging to keep up with the latest advances.
In the article, the author reviews the progress made in the field of remote sensing analysis and introduces the popular transformer-based methods in the field of remote sensing. The main contributions of the article are as follows:
Provides an overall overview of the application of transformer-based models in remote sensing imaging, and the author is the first to investigate the use of transformers in remote sensing analysis, bridging computer vision and remote sensing in this rapidly developing and popular field. the gap between recent advances in the field.
- Provide an overview of CNN and Transformer and discuss their respective advantages and disadvantages.
- More than 60 transformer-based research works in the literature are reviewed and the latest progress in the field of remote sensing is discussed.
- Discuss the different challenges and research directions of transformers in remote sensing analysis.
The rest of the article is organized: Section 2 discusses other related research on remote sensing imaging; Section 3 provides an overview of different imaging modes in remote sensing; Section 4 provides a brief overview of CNN and visual transformers; Section 5 reviews very high resolution (VHR) imaging; Section 6 introduces hyperspectral image analysis; Section 7 introduces the progress of transformer-based methods in synthetic aperture radar (SAR); Section 8 discusses future research directions .
Please refer to the original paper for more details.
- Paper link: https://arxiv.org/pdf/2209.01206.pdf
- GitHub address: https://github.com/VIROBO-15/Transformer-in-Remote-Sensing
The above is the detailed content of Reviewing more than 60 Transformer studies, one article summarizes the latest progress in the field of remote sensing. For more information, please follow other related articles on the PHP Chinese website!

1 前言在发布DALL·E的15个月后,OpenAI在今年春天带了续作DALL·E 2,以其更加惊艳的效果和丰富的可玩性迅速占领了各大AI社区的头条。近年来,随着生成对抗网络(GAN)、变分自编码器(VAE)、扩散模型(Diffusion models)的出现,深度学习已向世人展现其强大的图像生成能力;加上GPT-3、BERT等NLP模型的成功,人类正逐步打破文本和图像的信息界限。在DALL·E 2中,只需输入简单的文本(prompt),它就可以生成多张1024*1024的高清图像。这些图像甚至

“Making large models smaller”这是很多语言模型研究人员的学术追求,针对大模型昂贵的环境和训练成本,陈丹琦在智源大会青源学术年会上做了题为“Making large models smaller”的特邀报告。报告中重点提及了基于记忆增强的TRIME算法和基于粗细粒度联合剪枝和逐层蒸馏的CofiPruning算法。前者能够在不改变模型结构的基础上兼顾语言模型困惑度和检索速度方面的优势;而后者可以在保证下游任务准确度的同时实现更快的处理速度,具有更小的模型结构。陈丹琦 普

Wav2vec 2.0 [1],HuBERT [2] 和 WavLM [3] 等语音预训练模型,通过在多达上万小时的无标注语音数据(如 Libri-light )上的自监督学习,显著提升了自动语音识别(Automatic Speech Recognition, ASR),语音合成(Text-to-speech, TTS)和语音转换(Voice Conversation,VC)等语音下游任务的性能。然而这些模型都没有公开的中文版本,不便于应用在中文语音研究场景。 WenetSpeech [4] 是

由于复杂的注意力机制和模型设计,大多数现有的视觉 Transformer(ViT)在现实的工业部署场景中不能像卷积神经网络(CNN)那样高效地执行。这就带来了一个问题:视觉神经网络能否像 CNN 一样快速推断并像 ViT 一样强大?近期一些工作试图设计 CNN-Transformer 混合架构来解决这个问题,但这些工作的整体性能远不能令人满意。基于此,来自字节跳动的研究者提出了一种能在现实工业场景中有效部署的下一代视觉 Transformer——Next-ViT。从延迟 / 准确性权衡的角度看,

3月27号,Stability AI的创始人兼首席执行官Emad Mostaque在一条推文中宣布,Stable Diffusion XL 现已可用于公开测试。以下是一些事项:“XL”不是这个新的AI模型的官方名称。一旦发布稳定性AI公司的官方公告,名称将会更改。与先前版本相比,图像质量有所提高与先前版本相比,图像生成速度大大加快。示例图像让我们看看新旧AI模型在结果上的差异。Prompt: Luxury sports car with aerodynamic curves, shot in a

人工智能就是一个「拼财力」的行业,如果没有高性能计算设备,别说开发基础模型,就连微调模型都做不到。但如果只靠拼硬件,单靠当前计算性能的发展速度,迟早有一天无法满足日益膨胀的需求,所以还需要配套的软件来协调统筹计算能力,这时候就需要用到「智能计算」技术。最近,来自之江实验室、中国工程院、国防科技大学、浙江大学等多达十二个国内外研究机构共同发表了一篇论文,首次对智能计算领域进行了全面的调研,涵盖了理论基础、智能与计算的技术融合、重要应用、挑战和未来前景。论文链接:https://spj.scien

译者 | 李睿审校 | 孙淑娟近年来, Transformer 机器学习模型已经成为深度学习和深度神经网络技术进步的主要亮点之一。它主要用于自然语言处理中的高级应用。谷歌正在使用它来增强其搜索引擎结果。OpenAI 使用 Transformer 创建了著名的 GPT-2和 GPT-3模型。自从2017年首次亮相以来,Transformer 架构不断发展并扩展到多种不同的变体,从语言任务扩展到其他领域。它们已被用于时间序列预测。它们是 DeepMind 的蛋白质结构预测模型 AlphaFold

说起2010年南非世界杯的最大网红,一定非「章鱼保罗」莫属!这只位于德国海洋生物中心的神奇章鱼,不仅成功预测了德国队全部七场比赛的结果,还顺利地选出了最终的总冠军西班牙队。不幸的是,保罗已经永远地离开了我们,但它的「遗产」却在人们预测足球比赛结果的尝试中持续存在。在艾伦图灵研究所(The Alan Turing Institute),随着2022年卡塔尔世界杯的持续进行,三位研究员Nick Barlow、Jack Roberts和Ryan Chan决定用一种AI算法预测今年的冠军归属。预测模型图


Hot AI Tools

Undresser.AI Undress
AI-powered app for creating realistic nude photos

AI Clothes Remover
Online AI tool for removing clothes from photos.

Undress AI Tool
Undress images for free

Clothoff.io
AI clothes remover

AI Hentai Generator
Generate AI Hentai for free.

Hot Article

Hot Tools

PhpStorm Mac version
The latest (2018.2.1) professional PHP integrated development tool

Safe Exam Browser
Safe Exam Browser is a secure browser environment for taking online exams securely. This software turns any computer into a secure workstation. It controls access to any utility and prevents students from using unauthorized resources.

SublimeText3 English version
Recommended: Win version, supports code prompts!

Dreamweaver CS6
Visual web development tools

SublimeText3 Mac version
God-level code editing software (SublimeText3)
