We propose a novel human-drone interaction paradigm where a user directly\ninteracts with a drone to light-paint predefined patterns or letters through\nhand gestures. The user wears a glove which is equipped with an IMU sensor to\ndraw letters or patterns in the midair. The developed ML algorithm detects the\ndrawn pattern and the drone light-paints each pattern in midair in the real\ntime. The proposed classification model correctly predicts all of the input\ngestures. The DroneLight system can be applied in drone shows, advertisements,\ndistant communication through text or pattern, rescue, and etc. To our\nknowledge, it would be the world's first human-centric robotic system that\npeople can use to send messages based on light-painting over distant locations\n(drone-based instant messaging). Another unique application of the system would\nbe the development of vision-driven rescue system that reads light-painting by\nperson who is in distress and triggers rescue alarm.\n