Deep Learning Models for Gesture-controlled Drone Operation

Recently Unmanned Aerial Vehicles (UAVs) or Drones have gained enormous attention in applications like military, agriculture, industry, etc. One approach of controlling the operation of a drone is using hand gestures, which enables designing a low-cost system. However, the accuracy of such a system highly depends on the gesture recognition models. We can use a neural network-based gesture recognition model, which is a widely accepted image recognition scheme. In this work, we first design three deep neural network-based gesture recognition models: simple Convolutional Neural Networks (CNN), VGG-16, and ResNet-50 to uncover the best model for drone control. We evaluate the proposed models over our generated hand-gesture images in terms of their accuracy, precision, and complexity. The analysis reveals that each of the three models has its advantages and disadvantages while balancing between accuracy and complexity. For example, Simple CNN offers 92% accuracy on the testing set validation with the lowest validation loss compared to VGG-16 and ResNet-50. Thus, users can choose one of the proposed models to match their drone application.