Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

ConstScene: Dataset and Model for Advancing Robust Semantic Segmentation in Construction Environments

Maghsood Salimi, Mohammad Loni, Sara Afshar, Antonio Cicchetti, Marjan Sirjani

TL;DR

ConstScene addresses the need for robust perception in construction environments by introducing a large-scale, annotated semantic segmentation dataset that covers diverse weather, lighting, and lens-dirt conditions, plus monocular depth maps. The authors validate the dataset with baseline experiments using U-Net and SegFormer, achieving up to $mIoU$ of 82.27% on the original data (SegFormer-B5) and demonstrating that data augmentation improves performance. They show that adverse conditions challenge depth estimation, underscoring the importance of robust perception pipelines. The open-source dataset and code promote research on adversarial robustness and practical deployment for autonomous construction robotics.

Abstract

The increasing demand for autonomous machines in construction environments necessitates the development of robust object detection algorithms that can perform effectively across various weather and environmental conditions. This paper introduces a new semantic segmentation dataset specifically tailored for construction sites, taking into account the diverse challenges posed by adverse weather and environmental conditions. The dataset is designed to enhance the training and evaluation of object detection models, fostering their adaptability and reliability in real-world construction applications. Our dataset comprises annotated images captured under a wide range of different weather conditions, including but not limited to sunny days, rainy periods, foggy atmospheres, and low-light situations. Additionally, environmental factors such as the existence of dirt/mud on the camera lens are integrated into the dataset through actual captures and synthetic generation to simulate the complex conditions prevalent in construction sites. We also generate synthetic images of the annotations including precise semantic segmentation masks for various objects commonly found in construction environments, such as wheel loader machines, personnel, cars, and structural elements. To demonstrate the dataset's utility, we evaluate state-of-the-art object detection algorithms on our proposed benchmark. The results highlight the dataset's success in adversarial training models across diverse conditions, showcasing its efficacy compared to existing datasets that lack such environmental variability.

ConstScene: Dataset and Model for Advancing Robust Semantic Segmentation in Construction Environments

TL;DR

ConstScene addresses the need for robust perception in construction environments by introducing a large-scale, annotated semantic segmentation dataset that covers diverse weather, lighting, and lens-dirt conditions, plus monocular depth maps. The authors validate the dataset with baseline experiments using U-Net and SegFormer, achieving up to of 82.27% on the original data (SegFormer-B5) and demonstrating that data augmentation improves performance. They show that adverse conditions challenge depth estimation, underscoring the importance of robust perception pipelines. The open-source dataset and code promote research on adversarial robustness and practical deployment for autonomous construction robotics.

Abstract

The increasing demand for autonomous machines in construction environments necessitates the development of robust object detection algorithms that can perform effectively across various weather and environmental conditions. This paper introduces a new semantic segmentation dataset specifically tailored for construction sites, taking into account the diverse challenges posed by adverse weather and environmental conditions. The dataset is designed to enhance the training and evaluation of object detection models, fostering their adaptability and reliability in real-world construction applications. Our dataset comprises annotated images captured under a wide range of different weather conditions, including but not limited to sunny days, rainy periods, foggy atmospheres, and low-light situations. Additionally, environmental factors such as the existence of dirt/mud on the camera lens are integrated into the dataset through actual captures and synthetic generation to simulate the complex conditions prevalent in construction sites. We also generate synthetic images of the annotations including precise semantic segmentation masks for various objects commonly found in construction environments, such as wheel loader machines, personnel, cars, and structural elements. To demonstrate the dataset's utility, we evaluate state-of-the-art object detection algorithms on our proposed benchmark. The results highlight the dataset's success in adversarial training models across diverse conditions, showcasing its efficacy compared to existing datasets that lack such environmental variability.
Paper Structure (15 sections, 3 figures, 6 tables)

This paper contains 15 sections, 3 figures, 6 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Data Collection and Preparation
  4. Data Collection and Sensor Setup
  5. Data Annotation
  6. Data Augmentation
  7. Depth Estimation
  8. Dataset Statistics
  9. Experimental Results
  10. Evaluation Metrics
  11. Experimental Setup
  12. Prediction Results
  13. Advantages of $\texttt{ConstScene}$ in Defending Against Adversarial Attacks
  14. Conclusion and Future Work
  15. Acknowledgements

Figures (3)

  • Figure 1: (a) Original image. (b) Overlay mask of the original image. (c) Augmented image. (d) Overlay mask of the augmented image.
  • Figure 2: Visual representation of samples from our dataset under various weather conditions, accompanied by the corresponding disparity map. The red circles indicate the disparity region with low-confidence matches.
  • Figure 3: Distribution of images in each class, including background (BG), Car, wheel-loader (WL), crusher, human, pile, and road.