[Lesson1] Improve model accuracy

Hi, i just completed the first lesson and created my image classification model. It consists of 4 classes, which are German auto brands (vw, bmw, mercedes, audi). Every of them has about 500 images. I followed the instruction from course. It was a great start. But the accuracy of the classifier seems a little bit lower. (about 75%). And it was kind of overfitting. I share the Colab Notebook (https://colab.research.google.com/drive/1OLfyLBRPU44SwYhkP7JX3YQ0S1TXCm5U )
if anyone of you is interested to give some advice to improve performance. I will be very appreciated that.

I am by no means an expert but I would suggest the following…

Extend the training in some places to see if the network has more to learn in each section.

Try a smaller scale at the beginning say 112 by 112.

It may be too hard to predict the car brand from a picture of a dashboard that comes from marketing material. I would have a look through the images and make sure it is relevant to the prediction task.

1 Like

I have also considered about this limit. It will be difficult to distinguish car brands. Anyways, I will give another try.

When it comes to GPS it’s generally a great Idea to collect data from different brand’s interfaces and systems. Here’s one for audi https://update-navi.today/