mlsanityData to 7 Oct 2026

CrackNex: a Few-shot Low-light Crack Segmentation Model Based on Retinex Theory for UAV Inspections

arXiv 2403.03063 ↗First posted 5 Mar 2024cs.CVcited by 12

Not on the map. Why?

Abstract (arXiv)

Routine visual inspections of concrete structures are imperative for upholding the safety and integrity of critical infrastructure. Such visual inspections sometimes happen under low-light conditions, e.g., checking for bridge health. Crack segmentation under such conditions is challenging due to the poor contrast between cracks and their surroundings. However, most deep learning methods are designed for well-illuminated crack images and hence their performance drops dramatically in low-light scenes. In addition, conventional approaches require many annotated low-light crack images which is time-consuming. In this paper, we address these challenges by proposing CrackNex, a framework that utilizes reflectance information based on Retinex Theory to help the model learn a unified illumination-invariant representation. Furthermore, we utilize few-shot segmentation to solve the inefficient training data problem. In CrackNex, both a support prototype and a reflectance prototype are extracted from the support set. Then, a prototype fusion module is designed to integrate the features from both prototypes. CrackNex outperforms the SOTA methods on multiple datasets. Additionally, we present the first benchmark dataset, LCSD, for low-light crack segmentation. LCSD consists of 102 well-illuminated crack images and 41 low-light crack images. The dataset and code are available at https://github.com/zy1296/CrackNex.

Key facts. Cited by 12 papers in the map (data through 7 Oct 2026). Builds on 30 of its 60 references that are in the map. First posted on arXiv 5 Mar 2024.

Builds on

30 of 60 references.

References · 30

  1. 2015Deep Residual Learning for Image Recognitioncited by 51,809
  2. 2017Model-Agnostic Meta-Learning for Fast Adaptation of Deep Networkscited by 4,861
  3. 2017Rethinking Atrous Convolution for Semantic Image Segmentationcited by 2,154
  4. 2021Restormer: Efficient Transformer for High-Resolution Image Restorationcited by 1,618
  5. 2016Neural Architecture Search with Reinforcement Learningcited by 1,489
  6. 2018Deep Retinex Decomposition for Low-Light Enhancementcited by 720
  7. 2019PANet: Few-Shot Image Semantic Segmentation with Prototype Alignmentcited by 449
  8. 2017Few-Shot Learning with Graph Neural Networkscited by 324
  9. 2017Meta Networkscited by 268
  10. 2021Hypercorrelation Squeeze for Few-Shot Segmentationcited by 193
  11. 2021Rethinking BiSeNet For Real-time Semantic Segmentationcited by 124
  12. 2022Cost Aggregation with 4D Convolutional Swin Transformer for Few-Shot Segmentationcited by 114
18 more
  1. 2022Self-Support Few-Shot Semantic Segmentationcited by 96
  2. 2020Few-shot 3D Point Cloud Semantic Segmentationcited by 77
  3. 2019Meta-Learning of Neural Architectures for Few-Shot Learningcited by 73
  4. 2021Mining Latent Classes for Few-shot Segmentationcited by 69
  5. 2022LiDAR Distillation: Bridging the Beam-Induced Domain Gap for 3D Object Detectioncited by 54
  6. 2021Meta-Learning with Task-Adaptive Loss Function for Few-Shot Learningcited by 49
  7. 2020Few-shot Medical Image Segmentation using a Global Correlation Network with Discriminative Embeddingcited by 48
  8. 2021Domain Generalization on Medical Imaging Classification using Episodic Training with Task Augmentationcited by 33
  9. 2022PP-LiteSeg: A Superior Real-Time Semantic Segmentation Modelcited by 33
  10. 2022Interclass Prototype Relation for Few-Shot Segmentationcited by 25
  11. 2022SATS: Self-Attention Transfer for Continual Semantic Segmentationcited by 24
  12. 2022CrackSeg9k: A Collection and Benchmark for Crack Segmentation Datasets and Frameworkscited by 22
  13. 2023Learning to Estimate 6DoF Pose from Limited Data: A Few-Shot, Generalizable Approach using RGB Imagescited by 18
  14. 2024LanDA: Language-Guided Multi-Source Domain Adaptationcited by 13
  15. 2020Optimized Deep Encoder-Decoder Methods for Crack Segmentationcited by 10
  16. 2023Real-time High-Resolution Neural Network with Semantic Guidance for Crack Segmentationcited by 10
  17. 2021Topic Driven Adaptive Network for Cross-Domain Sentiment Classificationcited by 3
  18. 2021Pavement Distress Detection and Segmentation using YOLOv4 and DeepLabv3 on Pavements in the Philippinescited by 2

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2024: 4; 2025: 3; 2026: 5.

12 papers.

  1. 2024Endora: Video Generation Models as Endoscopy Simulatorscited by 53
  2. 2024Real-Time Pill Identification for the Visually Impaired Using Deep Learningcited by 20
  3. 2024Event-guided Low-light Video Semantic Segmentationcited by 14
  4. 2024Staircase Cascaded Fusion of Lightweight Local Pattern Recognition and Long-Range Dependencies for Structural Crack Segmentationcited by 6
  5. 2025Cognitive-Inspired Hierarchical Attention Fusion With Visual and Textual for Cross-Domain Sequential Recommendationcited by 2
  6. 2025BeltCrack: the First Sequential-image Industrial Conveyor Belt Crack Detection Dataset and Its Baseline with Triple-domain Feature Learningcited by 2
  7. 2025Deep Learning for Crack Detection: A Review of Learning Paradigms, Generalizability, and Datasetscited by 1
  8. 2026Prototype Learning-Based Few-Shot Segmentation for Low-Light Crack on Concrete Structurescited by 0
  9. 2026Robust Lightweight Crack Classification for Real-Time UAV Bridge Inspectioncited by 0
  10. 2026Multi-Task Crack Foundation Model for Engineering-Reliable Crack Representation and Topology Preservation in Civil Infrastructurecited by 0
  11. 2026Bridge Damage Detection from Low-Light UAV Imagery via Degradation-Aware Mixture-of-Experts Enhancementcited by 0
  12. 2026ReVA: A Scene-Centric Dataset Beyond Repetition for Remote Sensing Video Question Answeringcited by 0