An Efficient Framework for Animal Breeds Classification Using Semi-Supervised Learning and Multi-Part Convolutional Neural Network (MP-CNN)

Abstract

The automatic classification of animal images is an onerous task due to the challenging image conditions, especially when it comes to animal breeds. In this paper, we built a semi-supervised learning based Multi-part Convolutional Neural Network (MP-CNN) that classifies 35,992 animal images from ImageNet into 27 different classes of animals. The proposed model classifies the animals on both generic and fine-grained level. The animal breeds are accurately classified using Multi-part Convolutional Neural Network with a hybrid feature extraction framework of Fisher Vector based Stacked Autoencoder. Furthermore, with Semi-supervised learning based pseudo-labels, the model classifies new classes of unlabeled images too. Modified Hellinger Kernel classifier has been used to re-train the misclassified classes of animals and thereby improve the performance obtained from MP-CNN. The model has experimented with varied tasks to analyze its performance in each of the cases. The experimental results have proved that the coalesced approach of MP-CNN with pseudo-labels can accurately classify animal breeds and we have achieved an accuracy of 99.95% from the proposed model. INDEX TERMSFisher vector, inception-V3, modified hellinger Kernel classifier, multi-part based convolutional neural network, pseudo-labels, semi-supervised learning, stacked autoencoder.

YOLO-animal: an animal detection network based on improved YOLOv5 and attention mechanism

“…With respect to the detection of animals, a limited body of research has been devoted to addressing this paramount concern. [11] proposed a semi-supervised learning-based Multi-part CNN (MP CNN) for wildlife detection, which provides state-of-the-art performance. More specifically, for detecting endangered animals, an accurate real-time object detection framework WilDect-YOLO [1] has been proposed for detecting multi-class endangered wildlife species.…”

Critical Evaluation of Architectural Rudiments from the Perspective of Religion

“…1.Meena et.al [1] developed the Multi-part Convolutional Neural Network (MP-ORB), which uses semi-supervised learning to categorize 35,992 animal photos from ImageNet into 27 different animal classes. The suggested classification system divides the creatures into broad and specific categories.…”

Klasifikasi Bumbu Dapur Pasar Menggunakan Metode Deep Neural Network Berbasis Android

“…Pengolahan Citra Pengolahan citra (image processing) merupakan proses yang bertujuan untuk memanipulasi dan menganalisis citra dengan bantuan komputer. Beberapa contoh bidang kehidupan yang membutuhkan pengolahan citra digital diantaranya adalah bidang kesehatan: segmentasi untuk membedakan bagian -bagian sel darah, deteksi kerusakan organ tubuh, deteksi keberadaan [6]. Pengolahan citra dilakukan untuk melakukan transformasi suatu gambar/citra menjadi informasi yang dapat dipahami oleh manusia dengan menggunakan Teknik tertentu diawali dengan akuisisi citra atau proses menangkap citra analog sehingga diperoleh citra digital menggunakan kamera digital.…”

“…Seperti: Exposure, Contrast, Dynamic Range, Saturation, dan masih banyak lagi. Dengan memvisualisasikan histogram, kita dapat meningkatkan tampilan visual suatu citra dan juga dapat mengetahui jenis pemrosesan citra apa yang dapat diterapkan dengan membandingkan histogram suatu citra [6]. Gambar disimpan sebagai nilai piksel, setiap nilai piksel mewakili nilai intensitas warna.…”