Srinath Yasam1, Dr.S. Anu H Nair2, Dr.K.P. Sanal Kumar3

1Research scholar, Department of CSE, Annamalai University, Chidambaram, India

2Department of CSE, Annamalai University, Chidambaram, India (Deputed to WPT, Chennai)

3 PG Department of Computer Science, R. V. Government Arts College, Chengalpattu, India,,


          India is second only to China in global output of rice, wheat, sugarcane, cotton, groundnuts, and fruits and vegetables. Furthermore, it contributed 25% of the world’s pulses over the past decade and will continue to do so through 2019. About 58% of India’s population relies on agriculture for their income. In southern India, where 90 percent of the population lives, rice is the staple food for nearly all of its residents. Here The greatest difficulty in agriculture is increasing output per acre of land. In southern India, rice is a staple food. As a result of widespread use, there is a sizable demand for rice and rice-based products.        There are two main projects in our research: First, being able to recognise the various types Measure the potential for rice seeds to germinate. Our studies focus on four major types of rice that are grown by farmers in Tamil Nadu: 1) Andhra Ponni, 2) Pusa Ponni, 3) Vellari, and 4) Ponni. 2. AtchayaPonni Tiruchirappalli, Tamil Nadu, India is the source for both KO50 and IR20, which were obtained from the Tamil Nadu Agricultural University. Three colour features, thirteen morphological features, and eight textural features are extracted for a total of twenty-four.           In this study, we propose a system to predict whether or not a rice seed will germinate by using previously trained convolutional neural network models. The proposed system’s main goal is to equip any person with the ability to engage in agriculture through the use of a computer vision system. It’s useful for picking out seeds. To put it another way, it boosts efficiency. In addition, it gives those who want to start farming but lack the necessary experience a place to do so. This study provided a straightforward and cost-effective method for estimating the likelihood that four distinct types of rice seed—AtchayaPonni, AndhraPonni, KO50, and IR20—will germinate. Applying CNN with pre-trained models like Alexnet, Resnet, and inception v3.0 prediction, we obtained 89%, 83%, and 98.44% accuracy, respectively, in our experimental analysis.

Keywords : CNN, Seed germinate, Computer vision, image processing.