Skip to content
Hi-Kisan

Hi-Kisan

Where Prediction, Insight, Weather, and Prices Meet AI for Farmers.

Created on 7th April 2024

Hi-Kisan

Hi-Kisan

Where Prediction, Insight, Weather, and Prices Meet AI for Farmers.

The problem Hi-Kisan solves

Addressing Crop Diseases with Deep Learning

In India, a staggering 35 to 45 percent of crop yield is lost to diseases, amounting to a substantial loss of around 290 billion INR. Out of 30,000 plant diseases observed globally, a significant portion, around 5,000, are prevalent in India. Recognizing this challenge, we embarked on a mission to develop a deep learning model to combat this issue.

Utilizing Convolutional Neural Networks (CNNs), we achieved remarkable accuracies exceeding 95%. But what sets our solution apart from existing ones? While many platforms can identify crop diseases, our model goes a step further. Not only does it accurately identify diseases, but it also provides tailored solutions in both English and Hindi, ensuring accessibility to a wider audience.

Beyond addressing crop diseases, we recognized another critical need among farmers - weather forecasting. Farmers often struggle to anticipate weather conditions, such as whether it will be sunny, rainy, or cold. To bridge this gap, we incorporated interactive graphs powered by Plotly, offering insights into temperature variations and weather patterns.

Furthermore, the plight of farmers extends beyond crop diseases and weather uncertainties. Exploitation and misinformation often lead to dire consequences, including instances of suicide among farmers. To alleviate this, we established a connection with government databases to provide farmers with precise information about commodities relevant to their farming regions. Armed with this knowledge, farmers can make informed decisions, reducing vulnerability and empowering themselves economically.

Our endeavor is not just about technology; it's about making a tangible difference in the lives of those who feed our nation. By harnessing the power of deep learning, language accessibility, and data-driven insights, we strive to create a brighter and more secure future for our farmers.

Challenges we ran into

Ok so one of the major challeneges was training of the dataset, Since CPU always takes a lot of time for it. EVen switching to google colab didnt work unless I reduced the batch size;
Second major challenge was gemini api and its fine tuning to crop based activities. Thankfully, after so much dry run and testing we were able to solve this issue.
Another issue was translation button positoning paraller to text since streamlit didnt allow them to be placed in same line,
This was thankfully saved by dividing website into columns

Discussion

Builders also viewed

See more projects on Devfolio