Kivy based manual image classification helper

 Previous post: Raspberry Pi camera time-lapse

I currently have a Raspberry Pi camera set up near my window taking a picture of the weather outside my house. Next I have to classify these images to meaningful categories if I am to do anything with this data. This is going to be a very tedious task. So to make things easier for me I built myself a helper using Python and Kivy.

GitHub repo here

Design

For my classification I needed two major categories, time of day and weather. For each of these I selected 4 of the most appropriate sub categories.

• Time of day
• Morning
• Afternoon
• Dusk
• Night
• Weather (I live in Sri Lanka)
• Clear
• Rain
• Cloudy
• Fog

I used toggle buttons for these so I can quickly toggle to one state and I can keep pressing the submit button until something changes in the images.

Performance

So was this successful. In my opinion it was quite a bit helpful. To take it for a test, I downloaded about 250 images from the Raspberry Pi and I was able to classify these images within 1 minute.

So currently the Raspberry Pi has been taking pictures for 5 days. Which means there are about (30 x 24 x 5) 3600 pictures to classify already in the Pi. I am planning to keep it running for about 10 days (7200 pics). This would mean it will take me about half an hour to classify these images, which is not that bad.

If you need to see my code, you can find it on my GitHub here