Machine Learning Methods in Image Recognition

Abstract

In recent years, computer vision technology has played an important role in the field of wildlife monitoring. Machine learning technique can help to detect animals in the images collected in the field, so as to better understand the animal behaviour and the protection of key conserved animals. Machine learning algorithms can be trained iteratively on a large amount of image data to capture the features associated with the image labels, thus improving the recognition rate. The aim of this study is to classify animal images (including giant pandas, bears, lions and tigers) in the field background using machine learning methods.

This study is about evaluating the performance of various algorithms on animal image classification tasks. The algorithms that will be evaluated include the k-nearest neighbours algorithm, the decision tree algorithm, the random forest algorithm, the gradient boosting decision tree algorithm, and the support vector machine algorithm. The evaluation process involves clear model evaluation criteria as well as comparative analyses of model performance. For model evaluation, we used specifications such as accuracy, average precision, and other measures to comprehensively evaluate the performance of the models.

The results of the study demonstrate that the random forest and gradient boosting decision tree algorithms achieved high accuracy and precision in our tasks. However, the random forest algorithm demonstrated faster training and prediction speeds, making it more suitable for practical applications requiring the rapid processing of large datasets. This discovery offers an important reference for the future selection of machine learning models suitable for specific application scenarios.