• Ashiya
• Dr. Kirti Jain

Object detection is a fundamental task in computer vision, enabling machines to identify and locate objects in images across appli- cations such as surveillance, autonomous systems, and medical imaging. However, most existing approaches rely heavily on supervised learning and require large annotated datasets, which are expensive and time- consuming to obtain. To address this limitation, this paper proposes an unsupervised object detection framework that combines convolutional autoencoder-based feature learning with K-Means clustering. The model learns compact latent representations of input images and groups sim- ilar features to identify object regions without requiring labeled data. The proposed approach is evaluated on representative sample images, achieving reconstruction error values in the range of 0.015 to 0.018 and clustering variance between 0.22 and 0.25. The results demonstrate that the model can effectively capture meaningful visual structures and per- form approximate object localization in annotation-scarce environments. This work highlights the potential of unsupervised learning as a practical alternative for object detection when labeled datasets are limited.

Unsupervised Learning, Object Detection, Clustering, Auto encoder, Feature Learning, Image Localization, Computer Vision.

"Performance Analysis of Unsupervised Object Localization Using Autoencoder and K-Means Clustering", International Journal of Emerging Technologies and Innovative Research (www.jetir.org), ISSN:2349-5162, Vol.13, Issue 10, page no.a902-a907, October-2026, Available :http://www.jetir.org/papers/JETIR2610092.pdf

"Performance Analysis of Unsupervised Object Localization Using Autoencoder and K-Means Clustering", International Journal of Emerging Technologies and Innovative Research (www.jetir.org | UGC and issn Approved), ISSN:2349-5162, Vol.13, Issue 10, page no. ppa902-a907, October-2026, Available at : http://www.jetir.org/papers/JETIR2610092.pdf