Kitchen Utensils Recognition using Fine Tuning and Transfer Learning
Stephen Karungaru · 2019
To support blind persons at home especially in the kitchen, this work proposes the recognition of kitchen utensils using video sunglasses. The recognition system is based on transfer learning/fine tuning an existing deep learning algorithms, VGG16. Initially, our system can recognize 6 kitchen items using 1354 images in 6 classes. The training/validation and evaluation sets are set at 80% and 20% respectively. Most of the training data was downloaded from the Internet. In this challenging and noisy data, we achieved and accuracy of 95% using the fine tuning learning.