HAND GESTURE CONTROLLED DRONE USING MACHINE LEARNING

Abstract

The objective of this project is to develop a hand-gesture controlled drone using , machine learning. This project uses the Convolution Neural Network model, combined with the background elimination to detect different hand gestures, it takesa running average of the background for 30 frames and afterward utilize that running normal to identify the hand that must be presented after the foundation has been appropriately perceived. A background elimination algorithm removes the hand picture from webcam and utilizes it to prepare to anticipate the kind ofsignal that is. After trials and errors, a suitable set oftraining parameters are defined and network structure that contains seven hidden convolution layers with “relu” as the activation function and one fully connected layer. The network is trained across fifty iterations with a batch size of sixty-four. The model achieves an accuracy of 94.4% on the testing results. The proportion of preparing set to approval set is 1000:100. The framework first continues with the pre-procedure ofthe caught picture by wiping out all the undesirable parts in a constant video succession utilizing OpenCV and Python. Separating, division, resizing and includes extraction are additionally acted all the while. Next, a feed forward procedure through the system is summoned to yield a yield lattice. In light ofthe yield network, the perceived motion can be resolved. This system is designed to customize the network for an individual user