#Introduction

Image processing employs mathematical approaches and procedures to improve, analyse, or alter digital images. Enhancement, restoration, segmentation, compression, and recognition are all components of image processing (Jain, 1988). Brightness, contrast, and colour harmony are used to enhance photographs. Restoring an image eliminates noise and restores clarity and detail. Image segmentation breaks an image into parts in order to identify and extract objects of interest. Encoding photos using fewer bits reduces their file size without diminishing their quality. Image recognition is the training of algorithms to identify certain objects, patterns, or attributes in an image, enabling visual data analysis and decision-making. Image processing is used in medical imaging, astronomy, robotics, and surveillance. Image processing is required for digital photography, video production, and computer graphics (Bovik, 2009).

#Purpose

An examination of the classification capabilities of image processing is the focus of this paper. This report's aims are to do pre-processing on picture data, extract image features, create image classifiers, and assess the models.

#Scope

The report focus on pre-processing on all of the images, which includes scaling the images to 256 by 256 and extracting features based on the average rgb value from each image (train, val, test). The features from the train dataset will be used for training the model, the features from the validation dataset will be used to tune the hyperparameter, and the features from the test dataset will be fed to the trained model so that it can make a prediction about the class that this image belongs to. Finally, the performance of the model will only be evaluated on the test dataset. The results that were acquired from the classifiers and models will be presented and analysed, suggestions will be made for the best models and classifiers based on the analysis, and recommendations will be offered on how to increase the efficiency of the classifiers and models. In this paper, the author made use of a variety of classifiers, including linear and non-linear ones, such as the Support vector machine, Random forest, Logistic regression, K-Nearest neighbour, and Convolutional Neural Network.

#Background Information

The author has been provided with a folder that is titled data weed classification examine classification capabilities. The folder includes a Val set, a Test set, and a Train set; for further information, see table 1. Table 1: Illustrate the information in the train, test, and val set of the data weed classification. DataSet Cleavers Charlock
Train 208 272
Val 90 58
Test 90 68

#Objectives

Create image classifiers as well as an access model by using Python to create a process for the processing of image data. This pipeline include: • Pre-processing, feature extraction, train classifiers with extracted features and labels from train, test, and val set. • Evaluate models with extracted features from test and val set with Visualisation. • Discuss and analyse result. • Present recommendations