We investigate the use of Quantum Approximate Optimization Algorithm (QAOA) for image segmentation and compare its performance against traditional Spectral Clustering. Both methods are applied to segment images into binary labels, using superpixel representations. We employ Adjusted Rand Index (ARI) and Normalized Mutual Information (NMI) to evaluate segmentation quality quantitatively. Our results show that QAOA consistently outperforms Spectral Clustering in terms of both metrics, highlighting its potential as a competitive approach for quantum-assisted image segmentation.