All sight

Inspiration

We waste a lot of produce to diseases and rodents when farming. This reduces the overall yield of a farm while still taking up a lot of farming space. During a pre-college course with Carnegie Mellon university I created an AI model which was able to identify diseases in different plant species. This was a simple project in its self however, it left me wondering what other useful information could we get out of simple pictures of plants.

What it does

All sight is an idea, its not a full app yet, but I will be further developing the app in the future. Apps for identifying plant diseases already exists, however, they do not fully utilize the power of image classification with AI.

All sight would not only identify diseases, but it would also identify deficiencies in the plant such as low nitrogen in the soil. This is easily identifiable by looking at the color of a plant's leaves. If the soil is lacking proper nutrients the leaves of the plant will begin to become pale and yellow due to the lack of chlorophyll. Moreover, the app would also identity what rodents and insect species may be affecting the plant not only based off pictures but also using location data to see what native species of animals and insects dwell in the area. The pictures of the plant would identify the bites and match them to other pictures of bites from animals and insects that are known to dwell in the area. The combination of image classification and location data is important because, if we were only to look at the bit itself we would have an extremely large database of insect/animal bites to match it to. With the addition of location data we can shorten the search which would hypothetically then increase the accuracy of the AI model.

Now, while identifying factors which may be affecting crops is important it still only solves half of the problem. Identifying proper treatment is the next challenge. All sight would recommend some remedies based off a google search similar to how Siri answers questions and some from the forum. The app would also host its own forum which would connect farmers to one another, which would provide a better solution to un seen problems since it implements each farmer that participated on the forum as a mind working together with others to solve a similar problem (think of the built in forum as a stack overflow for farmers). On top of this the app would recommend products which match the needs of the plant, and this is where the app would generate revenue. In order for products to be recommended the vendors must subscribe to the app. This offers vendors advertising and also helps farmers find a product that best suits their needs. This feature would be implemented later on in the app once it gains enough traction to have a reach which is attractive to vendors (say 10k active users).

Although there is no physical product yet I hope to further develop this in the future