FarmLens

• 

  The final Output of the program on the webapp

Inspiration:

As a son of a family of farmers I understood that farming is not just a livelihood but a way of life, every harvest is a gamble. Farmers wake each day to tend their fields with hope but also with fear. A single unseen pest a silent crop disease or a delay in treatment can wipe out months of hard work and push families into debt. This is where FarmLens was born.

What it does:

Imagine a farmer taking just a simple photo of his crop leaf with his phone. Instantly FarmLens, powered by AI scans the image, identifiesthe crop and potential diseases, and provides clear, actionable guidance: what treatment to use, what dosage, and how to prevent it from spreading. No waiting for distant experts. No wasting money on the wrong chemicals. Just timely, reliable advice right in the farmer’s hands.

How we built it:

We built a dataset of images using roboflow and clearly anotated the images with the diseases of few of the main diseases that occure in indian agriculture sector. Then trained the dataset to to make a datamodel for YOLOv8 that uses image detection to run and provide what type crop is being detected along with diseases are present on the leaves along with its preventive measures and cure for the current disease. After this we made a flask server and made a webapp for the same so that it was more accessible for people

Challenges:

We ran into a few challenges such as the making of dataset of repetative and tedious for just 2 teammates to handle. Then while training the datamodel for yolo we ran into some bugs in the dataset which reduced the confidence curve tremendously we found that there was a imbalance of images of one perticuler diseases so the model trained would not detect the diseases properly to fix that we added more images for the remaining diseases along with adding a few data pre-processes to make the authencity of the dataset better.

Accomplishments that we're proud of:

We were able to succesfully run the program and get promissing outputs and futher we were able to integrate a drone based system that would go and help farmers to click picture and send to the model to detect using a DJI tello

What we learned:

We learnt about diffrent diseases that are present on crops and what challenges are faced by our beloved farmers and how IOT based applications could help them minimise the massive responsibilty of providing food for the county.

What's next for FarmLens:

We are planning to grow FarmLens into a mobile application and futher more in the future we are planning to use IOT based devices to help automation of the farming process using sensors to detect water quality, soil quality etc.

Built With

• cv2
• drone
• flask
• java
• os
• python
• roboflow
• tellopy
• ultralytics
• yolo