Automatic detection using image processing techniques provides fast and accurate outcomes with the advancement in computer vision by presenting an opportunity to expand and boosts the practice of precise plant protection and extending the computer vision applications in the field of agriculture.
Usually, farmers or experts observe the plants with the naked eye for detection and identification of diseases in plant leaves which is time-consuming, expensive, and imprecise.
A cutting-edge approach has come into reality for the development of plant recognition models based on the image classification using the deep convolutional networks.
The fundamental steps required for implementing the disease recognition model starting from collecting images in order to create a database, assessed by the experts of agriculture, a deep learning framework to perform the deep CNN training.
A novel approach in detecting plant diseases using the deep convolutional neural network trained and fine-tuned to fit efficiently to the database of a plant’s leaves that was gathered individually for diverse plant diseases.
The advance and innovation of the developed model lie in its simplicity, healthy leaves and background images are in line with other classes, by enabling the model to distinguish between the ones that are diseased and healthy leaves using the deep convolutional neural network.
Handling common digital image processing techniques such as color analysis and thresholding used with the aim of detection and classification of plant diseases.
An automated system designed to help recognize the plant diseases by its appearance and visual symptoms which makes a great help to amateurs in the gardening process and for the professionals as a verification system in disease diagnostics.
In machine learning and cognitive sciences, the Artificial Neural Network is an information processing paradigm that is inspired by the way the biological nervous system such as the brain processes information.
Preprocessing of the images generally involves removing low-frequency background noise, by determining the intensity of the individual images, removing reflections and masking portions of images.
Training the deep convolutional neural network for making an image classification model from a dataset was proposed initially.
TensorFlow is an open-source software library for numerous computations using data flow graphs wherein the nodes in the graph represent the mathematical operations, while the graph edges represent the multidimensional data arrays as tensors communicate between them.
A convolution operation on small regions of input is introduced to diminish the number of free parameters. Rectified Linear Units are used as a substitute for saturating nonlinearities.
Nonetheless, as a result of extensive research deep learning techniques have shown better results in pattern recognition, image segmentation and object detection. Enormous tests were performed in order to check the performance of the newly created model.
An expansion of this study will be on gathering images for enriching the database by improving the accuracy of the model using distinct techniques of fine-tuning and augmentation.
The crucial goal for future work will be developing a complete system consisting of server-side components containing a trained model and an application for smart mobile devices with features such as displaying recognized diseases in fruits, vegetables, and other plants, based on leaf images captured by the mobile digital cameras.
Moreover, future work will involve widening out the usage of the model by training it for plant disease recognition on wider land areas, combining aerial photos of orchards snatched by drones, and convolution neural networks for object detection.
By extending this research, there is hope to achieve a valuable impact on sustainable development by enhancing crop quality for forthcoming generations.