Image
To prepare image files for training a model on Daysie, users must categorize images into their respective classes, name the folders according to these classes, and then compress all folders into a .zip file with a size not exceeding 1GB.
To prepare image data for model training on Daysie, follow these steps:
- Create folder name of classes (e.g., “Cat , Dog”).
- Place images into their respective subfolders. Ensure images are in .jpg, .png, or .jpeg format.
- Compress the main folder into a .zip file.
Audio
To construct an audio dataset for model training on Daysie, establish a hierarchical folder structure. Create a root folder and subfolders corresponding to each audio class. Subsequently, categorize audio files into their respective subfolders, ensuring they are in a supported format (e.g., .wav, .mp3). Once the dataset is organized, compress the root folder into a .zip archive.
To prepare audio data for model training on Daysie, follow these steps:
- Create folder name of classes (e.g., “OK , NG”).
- Place Audios into their respective subfolders. Ensure audios are in .mp3, .wav, or .mp4 format.
- Compress the main folder into a .zip file.
Time series (CSV)
To create a time series dataset for Daysie, a .csv file format is required. This file must include a dedicated column to represent the target variable or class label. Valid label formats include binary (e.g., 0/1), categorical (e.g., OK/NG), or numerical (e.g., Normal/Abnormal).
This .csv file can be used to upload data for model training on the Daysie platform
Create zip file
To create a ZIP file, right-click on the selected folder and choose ‘Compress to ZIP file
The output will be a ZIP file. This ZIP file is ready to be uploaded to Daysie for model training
