Claims:The invention, “AUTOMATIC MANGO LEAF DISEASE DETECTION USING DEEP CONVOLUTIONAL NEURAL NETWORK”
[1] Identifies and classifies bacterial, viral and fungal diseases in the plant leaves more effectively.
[2] Reduces the use of pesticides by early detection of the diseases and in turn provides healthy crop.
[3] Improves the crop production by early detection of the leaf diseases.
[4] Decreases the labor requirement and it’s a less time consuming process.
, Description:FIELD OF THE INVENTION
[01] This invention relates to the field of Computer Science and Engineering. Leaf diseases leads to production loss, which can be tackled with continuous monitoring. Manual leaf disease monitoring is tedious process and error-prone. Early detection of leaf diseases using deep learning can help to reduce the adverse effects of diseases and also overcome the shortcomings of continuous human monitoring. This invention, proposes a smart and efficient technique for the detection of leaf disease for mango trees which uses convolution neural network based deep learning classification model.

BACKGROUND OF THE INVENTION
[02] Agriculture is the backbone of India’s economy. Plants are a major source of food for the world population. In agricultural crops, leaf plays an important role to provide information about the amount and nature of horticultural yield. Several factors affect food production such as climate change, presence of weed, and soil infertility. Leaf disease is a global threat to the growth of several agricultural products and a source of economic losses. They are one of the underlying causes for the decrease in the number of quantity and quality of the farming crops. Manual plant disease monitoring is both laborious and error-prone. The failure to diagnose disease in plants leads to insufficient pesticide/fungicide use. Too much chemicals are dangerous for the crops as well as the human and farming land.
[03] The naked eye examination of a trained professional is the prime technique adopted in practice for plant disease detection. In order to achieve accurate plant disease diagnostics a plant pathologist should possess good observation skills so that one can identify characteristic symptoms. But it cannot produce fruitful result all the time. To address the above issues, it is necessary to detect the plant diseases by advanced and intelligent techniques. Some diseases do not have any visible symptoms, or the effect becomes noticeable too late to act, and in such situations, a sophisticated analysis is obligatory.

[04] In this invention automated techniques for leaf disease detection were introduced which helps the farmer in identifying the various diseases occurs in plant leaves. A novel deep neural network classification model is proposed for the identification of leaf disease using plant image data. The results produced in this kind of technique are more efficient for both small and large production of crops. Importantly the results are accurate and take very less time to detect the diseases.
PRIOR ART STATEMENT
1. CN111369540A - Plant leaf disease identification method based on mask convolutional neural network – 2020
2. AU2021101682A4 - Automatic plant leaf disease diagnosis with machine learning and deep convolutional neural networks - 2021
3. US10721859B2 - Monitoring and control implement for crop improvement - 2019
4. CN107798356A - Crop leaf disease recognition method based on depth convolutional neural networks – 2017
5. AR108473A1 - System and method for detecting plant diseases - 2017
6. WO2020086774A1 - Detecting infection of plant diseases with improved machine learning - 2019
7. KR20200087342A - Plant recognizing system based deep learning - 2018
8. CN108304812A - A kind of crop disease recognition methods based on convolutional neural networks and more video images - 2018
9. CA3073713A1 - Apparatus and method for agricultural data collection and agricultural operations - 2018
10. CN111400536A - Low-cost tomato leaf disease identification method based on lightweight deep neural network – 2020

REFERENCES
1. S.Suresh, R.Nagarajan, L.Sakthivel, V.Logesh, C.Mohandass and G.Tamilselvan, “Transmission Line Fault Monitoring and Identification System by Using Internet of Things,” International Journal of Advanced Engineering Research and Science (IJAERS), Vol - 4.Issue - 4, pp. 9-14, Apr- 2017.
2. S. Renuga, “An Identification of Variety of Leaf Diseases using DataminingTechniques”, IJCRCST –

International Journal of Contemporary Research inComputer Technology, (2017), pp. 100-102.

3. Kshyanaprava Panda Panigrahi, Himanshu Das, Abhaya Kumar Sahoo, Suresh Chandra Moharana, “Maize Leaf Disease Detection and Classification usingMachine Learning Algorithms”, (2020), pp. 659-669.
4. Abirami Devaraj, Karunya Rathan, Sarvepalli Jaahnavi and K Indira, “Identification of Plant Disease using

Image Processing Technique”, International Conference on Communication and Signal Processing

(ICCSP), Chennai, (2019), pp. 749-753.

OBJECTIVE OF THE INVENTION
[05] The main objectives of this invention are,
• To develop a system that has a capability to detect and classify the diseases in plants from their leaves.
• To improve the crop production by early detection of leaf diseases of mango tree.

SUMMARY OF THE INVENTION
[06] Agricultural is the backbone of any country’s economy. Leaf diseases in plants contribute to production loss, which can be tackled with continuous monitoring. The traditional method requires more labour work and it is time consuming. Leaf disease detection will be easier when it is visible to human’s naked eye. If the farmer has enough knowledge about the crop and continuous monitoring of the crops, then the disease will be detected and treated earlier. When the number of crops increased or the production is high then this type of detecting the leaf disease won’t work. The timely identification and early prevention of crop diseases are essential for improving crop yield and quality.
[07] It is observed that many farmers use pesticides to remove the effect of disease without confirming the specific diseases. The usage of pesticides affects the plant quality as well as human health. Due to climatic conditions, different types of diseases are affecting the crops. The different types of leaf diseases are bacterial disease, fungal disease and viral disease. The leaf spot is the common type of symptoms for bacterial disease. The symptoms for viral disease are mosaic leaf pattern, crinkled leaves and yellow coloured leaves. Fungal diseases are commonly found on wide range of vegetables. These diseases are responsible for enormous damage on plant.
[08] An automated technique for leaf disease detection makes the farmer’s life easier. In this invention, deep convolutional neural network models are implemented to identify and classify diseases in plants from their leaves. Deep Convolutional Neural Network is used to classify the diseased and healthy leaves and to detect the disease in the affected leaves. This technique is efficient for both small and large production of crops. Importantly the identification is very accurate and it takes very less time to detect the diseases.
[09] At first the image will be captured from the field with digital camera or smart phone. The image pre-processing techniques were applied to prepare acquired images for further analysis. After this, pre-processed images were inserted into the CNN algorithm to feature extraction with neural network. Then, best-suited extractions to represent the image are extracted from the image using an image analysis technique. Based on the extracted features, the training and testing data that are used to identify are extracted. Finally, a trained knowledge base classifies a new image into its class of syndrome

DETAILED DESCRIPTION OF DIAGRAMS
The description for Fig.1 is given below:
[10] Bacterial Disease: Bacteria are single-celled microscopic organisms. They used to attack living plants and cause plant diseases. Bacteria can be carried from one plant to another by wind, rain splash and insects. These diseases occur mainly on leaves, but some may also occur on stems and/or fruit.
[11] Viral Disease: Viruses are obligate parasites which require a living host for their growth and multiplication. They enters plant cell through plasmodesmata and to various plant parts by the phloem. The nucleic acid in plant viruses is the major infectious component of a virus, once the virus enters the plant cell they shed their protein coat and multiplies itself. Viruses will spread through infected seeds, grafting, wind, splashing and pollination.
[12] Fungal Disease: Fungi are the most common parasites causing plant disease. These are microscopic plants that feed on living green plants or on dead organic material. They can be carried from plant to plant by wind, water, insects and equipment. In order for fungus spores to begin new infections, adequate moisture and the right air temperature are required. Sometimes a plant wound is also needed for the fungus to enter the plant. Fungus diseases are common during wet, humid seasons.

Fig. 2 shows the flow diagram of the overall process
[13] Image Acquisition: In image acquisition the image will be captured from the digital camera or webcam. The captured image will be in either RGB or gray scale image.
[14] Image Pre-processing: After capturing the image, pre processing will be done on the image. The captured image can’t be used without pre-processing because there will be a lot of disturbances and noises will be present in the image. So in this step the noises from the image are required to be removed to obtain an accurate result.
[15] Feature extraction: Feature extraction helps to remove the redundant data from the data set. It is useful when we have a large data set and need to reduce the number of resources without losing any important or relevant information.
[16] Training CNN: Training a neural network consists of two phases. In forward phase, the input is passed completely through the network. In backward phase, gradients are back propagated and weights are updated.
[17] Convolutional Neural Network: A convolutional neural network is a feed-forward neural network that is generally used to analyze visual images by processing data with grid-like topology. A convolutional neural network is used to detect and classify objects in an image.
[18] In this invention, the CNN model is built to identify both healthy and diseased leaves. The images are trained and the output will be produced according to the input leaf. Finally, the decision will be taken whether the given input leaf is healthy or affected by plant diseases.