A geometric and deep learning reproducible pipeline for monitoring floating anthropogenic debris in urban rivers using in situ cameras

TL;DR

This paper introduces a comprehensive pipeline combining deep learning and geometric modeling to monitor and estimate the size of floating debris in urban rivers using in situ cameras, addressing environmental and operational challenges.

Contribution

It presents a novel framework integrating deep learning for debris detection and a geometric model for size estimation, tested under diverse environmental conditions and data biases.

Findings

Deep learning models vary in accuracy and speed depending on environmental conditions.

The dataset protocol, including negative images and temporal leakage considerations, is crucial for model performance.

Metric object size estimation using projective geometry is feasible and effective.

Abstract

The proliferation of floating anthropogenic debris in rivers has emerged as a pressing environmental concern, exerting a detrimental influence on biodiversity, water quality, and human activities such as navigation and recreation. The present study proposes a novel methodological framework for the monitoring the aforementioned waste, utilising fixed, in-situ cameras. This study provides two key contributions: (i) the continuous quantification and monitoring of floating debris using deep learning and (ii) the identification of the most suitable deep learning model in terms of accuracy and inference speed under complex environmental conditions. These models are tested in a range of environmental conditions and learning configurations, including experiments on biases related to data leakage. Furthermore, a geometric model is implemented to estimate the actual size of detected objects from a 2D image. This model takes advantage of both intrinsic and extrinsic characteristics of the camera. The findings of this study underscore the significance of the dataset constitution protocol, particularly with respect to the integration of negative images and the consideration of temporal leakage. In conclusion, the feasibility of metric object estimation using projective geometry coupled with regression corrections is demonstrated. This approach paves the way for the development of robust, low-cost, automated monitoring systems for urban aquatic environments.