Semantic Segmentation of Small Objects and Modeling of Uncertainty in Urban Remote Sensing Images Using Deep Convolutional Neural Networks

By Michael C. Kampffmeyer, Arnt-Børre Salberg, Robert Jenssen. 2016

572 citations recorded by the source corpus.

Subfields: Computer Vision and Pattern Recognition, Environmental Engineering, Media Technology, Remote Sensing

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Vision and Pattern Recognition

In Environmental Engineering

In Media Technology

In Remote Sensing

Citations in the corpus

Towards Safe Autonomous Driving: Capture Uncertainty in the Deep Neural Network For Lidar 3D Vehicle Detection (2018, methodology): “tainty in BNNs at test time by sampling the network multiple times with dropout. This dropout sampling method has been applied to active learning for cancer diagnosis [21], semantic segmentation [22]–[24] and image object detection for open-set problem [25]. III. NETWORK ARCHITECTURE Our proposed pr”
Concrete Dropout (2017, methodology): “We show improved accuracy and uncertainty on popular datasets in the ﬁeld, and further demonstrate our variant on large models used in the Computer Vision community, showing a signiﬁcant reduction in experiment time as well as improved model performance and uncertainty calibration.”
Building Extraction in Very High Resolution Remote Sensing Imagery Using Deep Learning and Guided Filters (2018, background): “Semantic segmentation of remote sensing imagery aims to classify every pixel into a given category, and it is an important task for understanding and inferring objects [9,10] and the relationships between spatial objects in a scene [11].”
Gated Convolutional Neural Network for Semantic Segmentation in High-Resolution Images (2017, background): “[29] focused on small object segmentation through measuring the uncertainty for DCNNs.”
Swin Transformer Embedding UNet for Remote Sensing Image Semantic Segmentation (2022, background): “Kammpffmeyer et al. [40] assembled patch-based pixel classiﬁcation and pixel-to-pixel segmentation, which introduced uncertain mapping to achieve high performance on small-scale objects.”
Beyond RGB: Very High Resolution Urban Remote Sensing With Multimodal Deep Networks (2017, methodology): “by using hallucination networks [45] to infer the missing modality as proposed in [46].”
Deep learning in remote sensing: a review (2017, background): “In [129], the authors studied the possibility of visualizing uncertainty of predictions (applying the model of [130]).”
Cross-City Matters: A Multimodal Remote Sensing Benchmark Dataset for Cross-City Semantic Segmentation using High-Resolution Domain Adaptation Networks (2023, background): “Kampffmeyer et al. [3] developed deep convolutional neural networks (CNNs) for semantic segmentation in terms of small objects in urban areas, where the uncertainty in CNNs is modeled by Bayesian approximation in Gaussian process [4].”
UNetFormer: A UNet-like transformer for efficient semantic segmentation of remote sensing urban scene imagery (2021, unknown): “However, within fine-resolution remotely sensed urban scene images, complicated patterns and human-made objects occur frequently (Kampffmeyer et al., 2016; Marcos et al., 2018) (Audebert et al.”
Weight Initialization Techniques for Deep Learning Algorithms in Remote Sensing: Recent Trends and Future Perspectives (2021, background): “[10] Semantic segmentation of small objects CNN He”
Multiattention Network for Semantic Segmentation of Fine-Resolution Remote Sensing Images (2020, methodology): “Hence, we normalize q i and k j by l 2 norm to ensure q Ti · k j ≥ − 1: We then rewrite equation 5 into equation 11, and simplify it into equation 12: ) The vectorized form of equation 12 is: As ) could be computed only once and reused for each query, time and space complexity of the linear…”
Miss Detection vs. False Alarm: Adversarial Learning for Small Object Segmentation in Infrared Images (2019, methodology): “Third, different from dense small object instance segmentation [18, 15, 2], the objects in ISOS are usually very sparse , e.g., containing a single instance only.”
Optical Remote Sensing Image Change Detection Based on Attention Mechanism and Image Difference (2021, background): “Many scholars have introduced deep learning technology to remote sensing processing and have achieved excellent performance in image classiﬁcation [25], [26], semantic segmentation [27]–[29], object detection [30], [31], change detection [17], [32]–[35], and other tasks.”
Classification With an Edge: Improving Semantic Image Segmentation with Boundary Detection (2016, result): “A similar comparison is reported in (Kampffmeyer et al., 2016), with a single upsampling layer, and also trained from scratch.”