Train Image Extraction

• 

• 

  A sample from the data file

Inspiration

We started off with HTML/CSS/PHP and found that in the end it gets boring really fast; so with 12 hours left we decided to start with the train API

What it does

It extracts the green box from an image of train tracks, saves it and then generates a string of numbers containing data such as the edges within the image, the pallet used, average colour. There is sample data included and in theory we can take the data stream and the name of each image and use something like SMAC to automatically recognise future images.

How we built it

We built it using Python3 along with numpy, cv2

Challenges we ran into

It is really hard to extract data from an image in a meaningful way.

Accomplishments that we're proud of

Despite being incredibly tired and starting late we actually managed to extract the images in a reasonable time and then build a picture.

What's next for Train Image Extraction

Use SMAC to train it to recognise the shapes, signs and name all of the images

Built With

• python3-and-hundreds-of-libraries