Image classification is now used to bridge the gap between computer vision and human vision, allowing machines to recognise images in the same way that humans do. It deals with assigning the appropriate class to the given image. We therefore propose a system and web application named AniDetNet using Efficientnet- B3 with transfer learning that classifies the given images as cows, cats, dogs, elephants, and pandas. This paper presents a modified network model and investigates the relationship between learning rate and accuracy before providing a computer application. As a result, the user gains knowledge about the given animal image while avoiding the risk of classification using CNN. The proposed methodology accounts 98.5% accuracy and recognition rate