Abstract:
The application of neural networks and image processing techniques for detecting using images dataset is presented in this paper. An image-based detection of diverse nuts types has been observed and shown using Deep Learning (DL). Artificially using image augmentation, additional training images generated by various processing methods or combinations of multiple processing, such as random rotation, moves, shearing and flipping. The DL model is developed based Convolutional Neural Networks (CNN) using Python in Google Colaboratory, or "Colab" platform. Using Google's environment provides a free access to GPUs as well as a few configurations is required. A dataset of 1595 images of four different classes of nuts were used for training, validating and testing in the model. The trained model reached an accuracy of 100% on testing set, representing viable approach in detection and classifications applications.
|