Creating a New Way to Classify Squash Varieties in India by Using CNN and LSTM Techniques

The study brings a new sorting system that uses Convolutional Neural Networks (CNN) and Long Short-Term Memory (LSTM). These help to correctly spot squash types common in India. The system has a big set of $\mathbf{1 2, 0 0 0}$ high-quality pictures and it’s good at recognizing them with an accuracy level of $\mathbf{9 5. 6 \%}$. The CNN part is very good at getting complex pictures. The LSTM model handles data in a sequence for different kinds of squash types. Accuracy, precision, recall, and $F 1$-score show how good the model is at making correct classifications. When comparing this method with the old ones, it shows how good it is. This proves that it could be helpful for farming activities. This shows how strong the system is at spotting small differences between squash types. It also means better protection of seeds, more effective farming methods, and meeting market rules. But, chances to make things better involve getting models more flexible with different weather situations and changes in genes. More studies on changing how the system works, using different types of information, and making it work anywhere will make the system bigger and more useful. Emphasizing how machine learning can change farming, it gives a new way to classify kinds of squash. Doing the task right is a big step forward for farmers and helping them make plant farming better in different parts of India.