Article Preview
TopIntroduction
There are various plant species on earth which provide food and shelter for the living beings. The environment on earth will not exist without plants. Worldwide, there are numerous species of plants which balance nature. Nowadays, earth faces various vital ecological problems like – global warming, loss of biodiversity and various environmental damages. At the very first step plants leaf are identified and classified on the basis of their shape, size, color and texture etc. On the other hand there are a number of ways to identify the plant such as fruit, root, flower, leaf, etc. leaf recognition is a very useful step in biodiversity.
Status Of Agriculture Stats In India And Global Scenario
Agriculture is a well known sector which plays an important role in the economy of the developing countries. Almost 47% of the geographical area has been occupied by the agriculture sector of India. More than 70% of the rural population depend on agriculture. In India, Cereal production occupies 15% of area, whereas pulses occupies 12% of area, and fruits or vegetables occupies area under 10%. Agriculture is the main source of food, income and employment for many people around the globe. An important evolution has been made by the conception of policy makers and economists regarding the role of agriculture in economic development. Around one fourth of the Earth’s surface is under cultivation, most of the land is converted for agriculture in the last 30 years. Many regions including Australia, Europe and recently Brazil and India are becoming really productive in terms of raising yields with the help of organic manures and fertilizers (Sharma, A. et al., 2018).
IOT, ML and Ai Use in Agriculture
Agriculture is considered as the backbone of Indian economy, as a large portion of the population of the country is dependent on it. Agricultural outcomes can be improved with the help of advancements made in the technological field. It has been reported that IoT, ML and AI can be used in order to enhance the quality of agriculture. These techniques have been the centre of attraction for researchers, so that they can apply these techniques in the field of agriculture.These latest techniques are being considered as the origin for modern agriculture. These techniques have the potential to make land more favourable for agriculture, as they can help in manifesting a model which takes decision by collecting data and information on various factors like temperature, wind speed, humidity, rainfall, content of soil, etc. The decisions made by this model can help in improving the quantitative as well as qualitative approaches. Also, in order to limit various risks and challenges, these techniques can be applied. Farmers may be able to address information and data for accurate fertilization programs(Bhatta, N. et al., 2019)(Garg, D. et al., 2020).
Literature Review
The shape of the leaf acts as a prominent feature in recognizing and classifying a plant. The basic geometry information which can be extracted from the shape of the leaf is the diameter, physiological length and width, area and perimeter of the leaf. The color of the leaf, texture and veins are also the features. To compare learning algorithms cross validation is used. In order to compare results across means, Linear Discriminative Analysis, Linear SVM and Quadratic SVM MATLAB is used. 2% improvement over the previous results was given by the optimized feature selection. It works for the model whose performance is best (Quadratic SVM with K-means background separation K=4), and has an optimized feature set. It was 93.1% accurate. For each classifier the F1 scores and Precision Recall are in the range of 85% to 100% (Singla, B. et al., 2019).
During the Making of the project we have learnt about many ways and processes for collecting, cleaning, applying different kinds of various machine learning models, and detecting the defects on various kinds of leaves. We have learnt about the various parameters of classifying and detection of different leaves. We learnt about various pre-processing techniques. We studied leaf feature extraction and its recognition approaches for classification of plants. Studied image binarization techniques like Otsu Threshold method. We studied various techniques for feature extraction like color extraction, shape extraction. We have studied various research papers related and useful to leaf identification techniques some of which are given in the below sections (Chawda, K. et al., 2015) (Jabal, Ab. Et al., 2013).