The purpose of the study is to analyze and compare the classification and identification results taken from various deep learning models and techniques used for computer vision 2D object classification tasks. Here for this study 6 different plant datasets were taken. Firstly, the work starts with CNN architecture from scratch with an accuracy result 73.1%. Again when it is trained with the augmented datas, the performance of the work gets increased as 82.88%. The aim of the current paper is to check the performance when pretrained Networks were applied. For that, using the same dataset one of the CNN model, ResNet50 were taken and after that this paper will check the performance and finally compares with other CNN model to classify the datasets and conclude the comparison results.