Tom Ordonez

Data Science, Machine Learning, Software Engineering

Stanford Dogs Dataset, Flutter, Dart, and Java

I thought of building an app about dogs, where the interface is swiping pictures of dogs.

A quick search for 'dog pictures dataset' turned to this result...

Stanford Dogs Dataset

Stanford Dogs Dataset

I downloaded these files:

  • images.tar
  • annotations.tar
  • lists.tar

The file lists.tar had these files:

  • file_list.mat
  • test_list.mat
  • train_list.mat

I created a new conda environment to open the mat files:

conda create --name dogs scipy

Loaded the data with:

data ='file_list.mat')

Here is a sample of the data:

'annotation_list': array([[array(['n02085620-Chihuahua/n02085620_10074'], dtype='<U35')],
   [array(['n02085620-Chihuahua/n02085620_10131'], dtype='<U35')],
   [array(['n02085620-Chihuahua/n02085620_10621'], dtype='<U35')],
   [array(['n02116738-African_hunting_dog/n02116738_9829'], dtype='<U44')],

I also opened the images.tar and found the files that corresponded to the data above:


This is the content of annotations.tar for the first file above:

            <database>ImageNet database</database>

From Flutter/Dart to Java

I studied Flutter and Dart for about a month. This is a great stack. However, I wanted to build knowledge that could be used for something other than just building an Android app. I am sure that Dart can be used for other types of applications. However, Java is still used for pretty much everything.

I dropped Flutter/Dart and decided to move forward with Java to build a few simple Android apps.