Double compression detection is one of the most popular techniques for the authentication of images and videos in the sphere of multimedia forensics. Several forensic methods have been suggested revealing video forgery, particularly high-efficiency video codec (HEVC) double compression detection. Even though forensic methods can effectively validate the integrity of videos, they become more intricate while practicing anti-forensic (AF) on manipulated videos. In this study, we present an innovative counter-anti-forensic (CAF) approach to detect anti-forensically misguided HEVC double compressed videos with the same coding parameters using coding pattern analysis. We mainly focus on employing generative adversarial networks (GANs) attacks to reconstruct artifacts that appear in compressed videos for deceiving forensic methods. In the proposed CAF method, we apply AF attacks twice in primary and secondary compressions and analyze the coding pattern changes to build a CAF method. We extract syntax information, and introduce new syntax map representations to improve distinguishability between single and double-compressed sequences. Additionally, we apply Gaussian and median filtering as AF attacks to show the robustness of the proposed method. The proposed deep learning-based CAF method shows better performance and can overwhelm state-of-the-art works.