Eyes on the Water – Thousands of citizen photos train AI to monitor river pollution

Using thousands of photos gathered by the Friends of Bradford’s Becks community group, Artificial Intelligence (AI) models have been trained to spot visual markers of river health. Merging WRc’s expertise in water quality with the AI skills of Rain++ and National Taiwan University, the Eyes on the Water project has demonstrated how photos – which can be easily and safely gathered by citizens – can provide insights into the health of our rivers.

Friends of Bradford’s Becks (FoBB) has over 1,000 members on social media, who all want the Bradford beck river system to thrive. They have collected around 100,000 photos of the streams that flow through Bradford – photos that can really help. In summer 2021, FoBB provided photos to Yorkshire Water as part of a ‘Pollution Watch’ project. This led to sewer misconnections being identified and subsequently fixed, helping to improve water quality.

River health monitoring is normally carried out with chemical and physical equipment by the Environment Agency and water industry professionals. This equipment is expensive and complicated to set up and maintain. Camera monitoring could offer a cheaper, more accessible way to monitor river health, supplementing traditional methods. Last year’s Big River Watch encouraged almost 6,000 people across the UK to take photos of their local river, illustrating widespread potential for this model to trigger actions to enable nature recovery, such as removal of obstructions or debris.

Analysing thousands of photos, however, is challenging. Through a six-month project funded by Natural England, we tested and trained AI models that would recognise signs of river health in photos. We started by defining visual markers of river health, including the presence of wildlife, outfalls discharging into the river, rubbish, obstructions to fish passage, channel modifications and discolouration. A subset of photos that contained these markers was manually labelled and then used to test and train AI models. Although application of image recognition models to environmental challenges is increasing ^[1], the dataset available for this study, links to river health, and transferability of project results are thought to be novel.

Different AI models were trialled during the project. We found that ChatGPT and C-Tran models performed similarly well, with C-Tran having the advantages of higher precision, being trainable and being free. Whilst YOLOv8 had lowest overall accuracy, its ability to draw bounding boxes around river health markers of interest was considered useful. The final models and supporting documentation are openly available via GitHub.

The final, trained model was used to see what citizen photos can tell us about the health of Bradford’s Becks. The most frequently identified markers were plants (99% of photos), discolouration (63%) and modified channel (28%), indicating the types of issues faced by Bradford’s Becks and of interest to FoBB members. Maps of these results suggest that channel modifications and discolouration occur throughout the Bradford Becks catchment – i.e. there are no hotspots. The analysis also identified outfall locations throughout the city, and a selection of these that were close to sewage litter. These outfalls could be a priority for further investigation by the water company or local authority.

Natural England senior advisor Dr Rachel Palfrey said “Clean water is essential for life. Natural England is delighted to support the Water Research Centre, Rain++, National Taiwan University and the Friends of Bradford’s Becks with this innovative project, using technology to help us protect, restore, and enhance water quality within the Bradford District. This initiative forms part of our wider Nature Recovery Project across Bradford and the South Pennines, where we are working collaboratively with partners to connect people with nature and support local habitat and landscape restoration that contributes to growing a national Nature Recovery Network.”

The project team has secured funding from Natural England to improve the AI models further in 2024.

In 2018, there was a heart-breaking, long lasting, dry weather, pollution incident which came from beneath the city. It killed much of the aquatic life downstream of the city centre. It was two years before a fish was spotted in the river again. In the aftermath of this devastating incident, members of FoBB thought about ways of remotely monitoring the river and to raise an alarm when pollution is spotted. This can easily be done with commercial, chemical and physical monitoring equipment such as that employed by the Environment Agency and water industry professionals. Such equipment is prohibitively expensive and beyond the means and practical skills of most community groups. It was thought that camera monitoring could offer a more affordable, practical alternative to the traditional methods. - Robert Hellawell, Friends of Bradford’s Becks, 2023.

^[1] Examples include Flora Incognita; Teng, Kylili and Hadjistassou, 2022; Koster seafloor observatory.