Designing and training a tiny CIFAR-10 model
The tight memory constraint on LM3S6965 forces us to design a model with extremely low memory utilization. In fact, the target microcontroller has four times less memory capacity than Arduino Nano.
Despite this challenging constraint, in this recipe, we will be leveraging the following tiny model for the CIFAR-10 image classification, capable of running on LM3S6965:
Figure 7.1 – A model tailored for CIFAR-10 dataset image classification
The preceding network will be designed with TF and the Keras API.
The following Colab file (in the Designing and training a tiny CIFAR-10 model section) contains the code referred to in this recipe:
prepare_model.ipynb
Getting ready
The network tailored in this recipe takes inspiration from the success of the MobileNet V1 on the ImageNet...
