In this paper, our purpose is to investigate the system architectures and thresholds to detect gestures for a gesture-based system. We developed two gesture recognition systems: DESigning In virtual Reality (DesIRe) and DRiving for disabled (DRive). DesIRe allows real-time dynamic interaction with an object to perform a number of design operations, controlling 1 LED and wearing Datagloves. DRive allows a quadriplegic person to control a car interface taking input from 2 LEDs and shoulder movements only, wearing a garment with 2 LEDs. We developed an interface using a stereoscopic display (VISOR) and various motion capture systems integrating Vizard Virtual Reality Toolkit, an optical tracking system, datagloves, and a sensor-based jacket. In this paper, we describe a generic system architecture for gesture detection and present an approach to represent the finger states using binary digits. We have found this approach is sufficient enough and have not detected any specific delays in the detection of gestures using different input systems.