<p>Recent advancements in robotic prosthetic hand technology have led to significant progress, resulting in the development of lighter-weight devices capable of implementing a diverse range of hand gestures to grasp and release objects. However, research on robotic prosthetic hand control continues to be actively pursued due to limitations in existing methods. Numerous methods rely on interpreting user’s intentions through diverse biometric signals, often requiring extensive training time and individualized education on usage methods, which can be time-consuming and challenging. To address these issues, we propose a real-time grasp control system that utilizes a wearable camera to capture 2D images and 3D depth information to guide the control of grasp and release actions. Furthermore, we present a technique that uses fingertip sensors for grasp control and demonstrate the feasibility of using wireless communication between a single-board computer and a deep learning server to leverage deep learning object detection. Grasp-and-release experiments showed a success rate of 80.2%. This method does not require customization for amputees and offers easy adaptability with minimal training, thereby increasing the accessibility and efficiency of prosthesis control.</p>