This paper presents a new design of an emotion detection model and a system for interpreting the human body language based on convolutional neural networks, which are used to retrieve features from images of humans, aiming for high precision and wide emotion detection ability. A large number of datasets of labeled body images is used for training. Then, a separate evaluation dataset is used. This will then be used it to detect any body parts and can be used to decipher meanings behind the languages presented by the body motion This futuristic system holds promise for various applications requiring emotional understanding. By analyzing intricate patterns and relationships within the data, the system learns to associate specific body language cues with distinct emotions. This complex learning process, driven by a classified large database of body images, enables the system to learn and get good accuracy in interpreting the hidden language.