Collect Ground Truth Data from FIFA/PES

I am a researcher working on Deep Learning. For a project I need extraction of ground truth data from video games especially from FIFA and/or PES. I found some papers/works that intercept DX11 calls and generate ground truth data from the games. One example is: [login to view URL] and the code: [login to view URL] I check the code and understand the overall idea. I want to apply the same idea for my project. I need the following ground truth labels for the project. The data should be taken from a soccer game FIFA/PES 2022. 1. Semantic segmentation data for different classes: player, referee, ball, pitch lines, goal post - already implemented in the code 2. Instance segmentation for different classes: player, referee, ball 3. Depth data - already implemented in the code 4. 3D bounding boxes for different classes wrt camera frame (3D object center point and width, height, depth of the object): player, referee, ball 5. 2D bounding boxes for different classes: player, referee, ball 6. Extraction of camera pose with respect to a defined coordinate frame (likely a coordinate system sit on the center point of the pitch) Some requirements are already fulfilled in the given code, some need to be developed. The code itself is extendible via shader injection so to fulfill given requirements additional shader code should be implemented and injected. In summary, I want someone run the code with the latest FIFA/PES 2022 and code additional DirectX shader code in order to collect mentioned ground truth data.