Pipeline 1.0 = Face detection (object detection model) -> classifier(a CNN or ViT) OR MediaPipe -> target classes : happy, sad,surprise, neutral.
Pipeline 2.0 = Classes -> robot sensors -> response according to the target classes
Pipeline 3.0 (optional) = Detects the t-shirt logos and then uses OCR and then tells some facts about it.