Concerning scenario of water pollution in todays times

Water is one of the most vital life resources. As the world population grows, the demand for potable water is ever increasing. At the same time, due to lack of organisation and awareness, we are polluting our very limited water resources like our rivers, lakes, ponds etc. It is said that the next world war will be fought for potable water. There is an SOS level need to preserve, conserve and save the water available and prevent it from getting polluted.

India's primary source of water is rainfall and due to the changing climate, rains are becoming increasingly erratic and hence preserving water is a crucial task. Although there have been efforts in this direction from the government but they seem to be a drop in the ocean because of how challenging this task is. These efforts are unable to battle the increasing levels of pollution.

For example, by the time plans to remove 1 ton of plastic from the river are implemented, there is already 5 times more plastic in the river. The problem is grave as the pollution level in the water bodies is rising at an alarming rate.

Currently, the pollution in rivers and lakes is noticed only when the pollution level goes to an irreversible stage and only war level efforts can reverse the change. To avoid this, we need to check it much beforehand. If we can check it on a daily basis and slight changes can be made noticeable, it can lead to saving of water resources.

In another scenario, rivers in the monsoon season get flooded due to heavy and torrential rains, which brings down a lot of silt. When there are floods in the catchment areas, the silt starts coming down and gets mixed with river water which is supplied to cities after treatment in water filtration plants. The increase in the level of silt goes unnoticed until it chokes the water filters and the city water supply gets disrupted due to the same. In order to check this, if we can take help from an on-site camera and keep a check on the silted water, the situation can be avoided.

How does this machine learning application help solve the problem?

With the help of on-site cameras placed near lakes, rivers and other such water bodies (i.e., required or suitable locations) pictures of the water resources can be taken at regular intervals, the frequency of which can be customised as per the season or levels of rains in the monsoon.

The model can be trained to recognise the pictures of water bodies as Polluted, Silted or Clean. As the model is fed with images of the water bodies taken by the on-site cameras, it will start classifying the images into one of the three categories as mentioned above. If the classification is polluted or silted for more than a predefined number of times, the system can be made to raise an alert and then timely action can be taken.

How can this model be taken be taken a step further?

This model can be enhanced further to recognise and classify high pollution, toxic foam pollution, oil spills, low-level pollution, rising pollution, rising silt etc. as it is fed with more real time images and more data is collected. This model can learn and improve over time and can slowly be rolled out to sites like ponds, lakes, creeks and backwaters in coastal areas.

Idea from personal experience