[Paper Review] Intelligent Debris Mass Estimation Model for Autonomous Underwater Vehicle

Mohana Sri S, S Swethaa+2 more|arXiv (Cornell University)|Sep 19, 2023

Water Quality Monitoring TechnologiesEnvironmental Science3 citations

TL;DR

This paper proposes an intelligent debris mass estimation model for autonomous underwater vehicles (AUVs) using YOLOv7-based instance segmentation to detect and segment marine debris from underwater images. By applying binary masking, morphological post-processing, and Monte Carlo area estimation, the method achieves accurate total debris area and mass estimation, enhancing AUVs' capability for environmental cleanup missions.

ABSTRACT

Marine debris poses a significant threat to the survival of marine wildlife, often leading to entanglement and starvation, ultimately resulting in death. Therefore, removing debris from the ocean is crucial to restore the natural balance and allow marine life to thrive. Instance segmentation is an advanced form of object detection that identifies objects and precisely locates and separates them, making it an essential tool for autonomous underwater vehicles (AUVs) to navigate and interact with their underwater environment effectively. AUVs use image segmentation to analyze images captured by their cameras to navigate underwater environments. In this paper, we use instance segmentation to calculate the area of individual objects within an image, we use YOLOV7 in Roboflow to generate a set of bounding boxes for each object in the image with a class label and a confidence score for every detection. A segmentation mask is then created for each object by applying a binary mask to the object's bounding box. The masks are generated by applying a binary threshold to the output of a convolutional neural network trained to segment objects from the background. Finally, refining the segmentation mask for each object is done by applying post-processing techniques such as morphological operations and contour detection, to improve the accuracy and quality of the mask. The process of estimating the area of instance segmentation involves calculating the area of each segmented instance separately and then summing up the areas of all instances to obtain the total area. The calculation is carried out using standard formulas based on the shape of the object, such as rectangles and circles. In cases where the object is complex, the Monte Carlo method is used to estimate the area. This method provides a higher degree of accuracy than traditional methods, especially when using a large number of samples.

Motivation & Objective

• To develop an intelligent model that enables autonomous underwater vehicles (AUVs) to estimate debris mass in marine environments.
• To address the challenge of accurate debris detection and segmentation in complex underwater scenes with low visibility and clutter.
• To improve AUV navigation and intervention capabilities by enabling precise estimation of debris volume and mass.
• To leverage deep learning and image processing techniques for real-time debris assessment in underwater operations.
• To support marine conservation by providing a scalable, automated solution for debris mass estimation in ocean cleanup missions.

Proposed method

• Utilizes YOLOv7 in Roboflow to generate object detection bounding boxes with class labels and confidence scores for marine debris in underwater images.
• Applies a convolutional neural network to produce segmentation masks by performing binary thresholding on the model's output.
• Refines segmentation masks using morphological operations and contour detection to enhance accuracy and reduce noise.
• Calculates the area of each segmented debris instance individually using geometric formulas for simple shapes (e.g., rectangles, circles).
• Employs the Monte Carlo method for area estimation of complex-shaped debris, improving accuracy with higher sample counts.
• Aggregates individual instance areas to compute the total debris area, which is then used to estimate mass based on assumed density.

Experimental results

Research questions

• RQ1 How accurately can instance segmentation with YOLOv7 detect and segment marine debris in real underwater imagery?
• RQ2 To what extent do post-processing techniques like morphological operations and contour detection improve segmentation mask quality?
• RQ3 How does the Monte Carlo method compare to geometric formulas in estimating the area of irregularly shaped debris?
• RQ4 Can the integrated model reliably estimate total debris mass from image data for AUV-based ocean cleanup missions?
• RQ5 What is the impact of image quality and environmental conditions on the performance of the debris mass estimation pipeline?

Key findings

• The integration of YOLOv7 with instance segmentation enables high-precision detection and localization of marine debris in underwater images.
• Post-processing techniques such as morphological operations significantly improved mask quality by reducing noise and filling gaps in segmented regions.
• The Monte Carlo method provided more accurate area estimation for complex-shaped debris compared to geometric approximation, especially with high sampling rates.
• The total debris area was computed by summing individual instance areas, forming a reliable basis for mass estimation.
• The proposed model demonstrates feasibility for real-time debris mass estimation in AUV applications, supporting autonomous environmental monitoring and cleanup.
• The method shows robustness in handling diverse debris shapes and underwater visual conditions, enhancing AUV operational autonomy.