ColorNet is a machine learning model developed at Clemson University that color corrects targeted portions of live video feeds in real-time. The original ColorNet 1.0 was developed to solve the problem of Clemson orange appearing incorrectly during football broadcasts. The machine learning algorithm can detect which pixels on the screen need color adjustment and shift those pixels to the correct color without noticeable latency in the broadcast and without negatively impacting surrounding colors. Production beta tests proved successful at correcting Clemson orange during live sporting events. ColorNet 2.0 was proposed as an improved solution that could correct any specified team color using image segmentation. The goal of ColorNet 2.0 is to use a reference color as an input and segment out portions of the screen containing that color. Then, the technician can adjust those areas of the screen according to the desired color specifications. This new approach would allow the model to be more universally applicable because it does not require new training data for each color the user wants to correct. It would also allow technicians more control over the final appearance of the brand colors. The current work presents a novel data augmentation strategy that synthetically expands the available training data to include all ACC colors. Future work will focus on the development of the neural network architectures needed to accurately automate the segmentation of the targeted brand colors.