Note: This document is only compatible with aid v0.x only. We are working on it for v1.0 and higher
When you are using AID for training, AID requires you to specify the file path of the training, validation and test dataset. In this chapter, we will introduct how the dataset should be organized.
You dataset is a zip file (Do not use .rar or other format since it's not free format). You can also use .7z file.
Inside the file, your dataset should looks like:
annotations.json is required for cvpm to understand your dataset. After the uncompression of your zip file, cvpm will check if this file exists, if it does not exist, cvpm will mark it as warning.
annotations.json is a list which looks like:
Since there are many different types of computer vision tasks, we hereby define several common data formats for the task.