Amazon: Exploring image file metadata and pixel-level features to uncover patterns, insights, and key questions.
1. Data frame
The purpose of the task is to test your ability to process image files and your knowledge of data-framing methods in a programming language of your choice. Based on the loaded photos, you need to create a data frame (e.g. Pandas DataFrame) with the following columns:
- should be replaced with a space)*The upper left corner of the image corresponds to the coordinates (0,0). If there are many brightest pixels, we choose the one closest according to the Euclidean norm to (0,0).
After creating the data frame, save it to images.csv file.
2. Aggregation of images
The second task is to sort the photos into subfolders. You need to sort the images by median brightness, divide them into bins of count 4, and save the images into subfolders with the naming scheme - [bin number]-images, numbering the bins from 1.
Attached to the task is a directory images, which contains image files with the following file naming scheme - stock-photo-[image content description]-[image id].jpg. The images should be loaded into any programming environment in the RGB color model.
Make sure that the solution reflects your entire thought process - it is more important how the code is structured rather than the final files.
See the full interview round by round, with questions for each stage.