Nilvana Vision Studio - Quick Start Guide
Please follow the instructions to get started with nilvana vision studio.
Create Administrator Account
Welcome to Nilvana Vision Studio. Before using for the first time, you need to create an administrator account. Keep the account name and password properly.
Set Up Mail Server
Only the administrator account can set up the mail server. Go to the email host setting page, fill in the information field, press "Save", and then click "Active" to start the mail server. Refer to "Nilvana Vision Studio - How To configure Gmail SMTP" for more information.
To create a dataset, enter the dataset name and type on the dataset overview page.
Upload Image Data
Click the dataset to go to the dataset information page and press "Upload images" to upload the image data to be annotated. Refer to "Nilvana Vision Studio - Dataset Management" for more information.
For the first annotation, we recommend using semi-automatic machine annotation and selecting COCO, the system’s built-in model to experience the convenience of accelerated annotation.
After the machine annotation is complete, click the "Manual annotation" button to manually check and confirm the results of machine annotation. Semi-automatic annotation will be represented by a dotted line. After confirming the annotation result, double left-click to recognize the annotation result.
Machine annotation is an auxiliary tool that can complete most annotation work. In the manual inspection process, the objects on any unannotated images can be annotated by manual annotation. After all annotations are complete, click "Leave" to return to the dataset information page.
Generate Dataset Version
In the dataset information page, switch to the version record page and click "Generate" to generate the dataset version. After becoming more familiar with Nilvana Vision Studio, you can use the data preprocessing items and data augmentation items to create more experimental combinations. Refer to "Nilvana Vision Studio - Dataset Versioning" for more information.
Press the "Training" button in the dataset version to view the model training status in the training plans page and wait for the model training to be completed. Refer to "Nilvana Vision Studio - Model Training" for more information.
You can view the trained models in the Models section and evaluate the results of model evaluation on the Model Evaluation page.
Congratulations! Your first computer vision model has been born.