RL-Ad: Integrating Eye-Tracking into a Reinforcement Learning Model for VR Video Ad Implantation
Qian Zhang, Yahui Tang, Chen Wang, Xun Zhang, Xiaoming Chen, Yuk Ying Chung · ACM Transactions on Sensor Networks · 2025
The implantation of advertisement (ad) within VR videos presents a meaningful opportunity to engage users in immersive and interactive ways, enhancing the potential for effective advertising. However, the challenge lies in seamlessly implanting ad while maintaining the user’s immersive experience. To solve this problem, we construct a dataset comprising VR images and videos with eye tracking data, providing a crucial resource for gaining in-depth insights into user interactions within virtual environments. Based on this dataset, we propose a reinforcement learning model called RL-Ad, built on a Deep Q-Network (DQN), which can dynamically adjust ad implantation regions according to users’ eye movements. The proposed RL-Ad model aims at more precisely capturing user attention and increase their viewing duration on the implanted ads. Experimental results demonstrate that the proposed model can significantly improve the ad implantation by optimizing ad position and size to capture user attention while avoiding the interference in user experience.