Our primary objective, was to build an image classifier on the CIFAR - 10 dataset, which worked on MobileNets, an efficient and recently developed architecture which has significantly lesser training time for a minute tradeoff in accuracy of the classifier, which is targeted at applications in embedded or mobile devices. We built the image classifier using a Convolutional Neural Network combined with a Feedforward Neural Network for classification purposes.
| Width Multiplier | Accuracy achieved(%) |
|---|---|
| 1.0 MobileNet | 71.2 |
| 0.8 MobileNet | 69.2 |
| 0.6 MobileNet | 70.7 |
| 0.4 MobileNet | 68.3 |
| 0.2 MobileNet | 68.1 |
We were able to build an artificial convolutional neural network that can recognize images with an accuracy of 71.2% using TensorFlow. We did so by pre-processing the images to make the model more generic, split the dataset into a number of batches and finally build and train the model.