Video-based person re-identification (ReID) remains a challenging problem due to variations in appearance, pose, and illumination across different camera views. Although models trained on large, fully labeled datasets achieve high accuracy, they often struggle to generalize effectively when applied to novel datasets. In this paper, we propose a novel approach using MiniMax Optimization Adversarial Network (MMOAN) to learn robust and domain-invariant feature representations for video ReID. Our method also integrates transformer-based feature extraction with adversarial learning to address both intra- and inter-domain shifts, improving model generalization across diverse datasets and camera configurations. We demonstrate the effectiveness of our approach through extensive experiments on three challenging video-based ReID datasets, where it achieves state-of-the-art performance, outperforming existing methods in terms of accuracy. The proposed method not only enhances the robustness of video-based ReID but also offers a scalable solution to domain adaptation challenges in surveillance networks.