The speaker performed rigorous processing of facial expressions from episodes with 1080p video by extracting faces, tracking pupils, and detecting blinks to understand the speaker's stress level without relying on heat maps, which vary from person to person. Complications arise with makeup, eyelashes, and obscurity from items such as baseball caps.