Disease Detection in Fruits Using Deep Learning Raju Hosakoti 1 , Soma Pavan Kumar 2 , Padmaja Jain 3 1 Student, Dept. of ECE, BNMIT, Bangalore, 2 Student, Dept. of ECE, BNMIT, Bangalore, 3 Assistant Professor, Dept. of ECE, BNMIT 1 raju17ec061@bnmit.in, 2 soma17ec079@bnmit.in, 3 padmaja.jain@bnmit.in Abstract— Crop cultivation plays an essential role in the agricultural field. Presently, the loss of crop is mainly due to infected crops, which reduces the production rate. It is very difficult to monitor the diseases manually. It requires a tremendous amount of work, expertise and excessive processing time. Hence deep learning used for the detection of diseases with more accuracy. This paper aims at providing a cost-effective and real-time solution to detect fruit diseases. CNN is used for feature extraction and classification. Deep learning concepts will help to identify the diseases in fruits (Apple) with more accuracy. Hence this results in predicting the disease in early stage so that the necessary actions to cure them can be taken immediately. Keywords— Convolutional Neural Network (CNN), Deep learning, Flyspeck, Sooty blotch, Apple Rot I. INTRODUCTION The emerging of new technologies such as digital image processing and image analysis technology has many applications in the biological field. In India, around 78% of the farmers are marginal and hence they are poor in resources. Therefore, they are not in a position to use the available resources for increasing productivity. A user-friendly software will help the farmer to some extent to detect whether the fruit is diseased or not. The image is processed using the Image Processing techniques and the disease is detected. The disease is detected by our image processing software that helps the farmers to take some precautions. This proves benefits in monitoring large orchards of fruits, and thus automatically detects the diseases as soon as they appear on fruits. The largest sector in the Indian economy is agriculture and it is also the largest employer. So, the agricultural industry places a significant role in increasing the Indian economy. Daily products in India are cheaper as compared to other parts of the world for example 1] In Europe 1kg of milk costs € 3.54 (as in Jan. 2021), around 313 rupees, whereas the same quantity of milk costs 46 rupees in India, which is 6.8 times lesser. 2] In Western Europe 1kg of apple costs € 2.5 (as in Jan. 2021), around 220 rupees, whereas the same quantity of apple costs 100rs in India. So, this is the average price variation between Milk, Fruits, Vegetables and many other agricultural products like Rice, Wheat compared to other countries. Hence there is a lot of scope for export, which is a new way for the farmer to earn more money which in term will increase the agricultural income which in turn increase the Indian economy. This is possible only if we produce more fresh fruits. So, this paper focuses on disease detection in fruits which helps in increasing the fresh fruits produced. Fruits like Apple, Grapes and Mango are called cash crops, hence improvement in productivity of these types of fruits is required. Various diseases may attack apple fruits mainly Flyspeck, Apple Rot, Sooty Blotch etc. So, its early detection is very essential. Here we use the deep learning concepts that make use of various CNN algorithms to improve the accuracy. In deep learning, we will train the images hence that helps for early detection of diseases. Fig 1. Example of Diseased Apple Fruit Fig1 shows the different types of diseases that may affect apple. The apple in the above figure is affected by 3 types of diseases they are, 1) Flyspeck 2) Sooty blotch 3) Scab. Flyspeck is a fungal disease with small dots on fruit, Sooty blotch will appear on the surface of the fruit and turns the fruit light black in colour, Scab will turn the fruit dark black. II. CONVOLUTION NEURAL NETWORK A Convolutional Neural Network is a Deep Learning algorithm that can take in an input image, assign importance to various aspects in the image and be able to differentiate one from the other. It is a multilayered neural network with a special architecture to detect complex features in data. CNN's have been used in image recognition, powering vision in robots, and for self-driving vehicles. Convolutional Neural Networks will make use of small pre-processing as compared to other image classification algorithms. Hence, CNNs can have the ability to optimization of filters (kernels). Journal of University of Shanghai for Science and Technology ISSN: 1007-6735 Volume 23, Issue 7, July - 2021 Page -309