Pre-trained Model
🤖 Guide to running models in the Models Center.
We recommend running pre-trained models in the Models Center to enhance annotation efficiency and precision. The article will show you how to do this.
Integration of custom models is now supported.
1. Select a Model
You can choose a model based on your data type and task requirements. Currently, BasicAI provides the following models:
Model | Data Type | Scenarios |
---|---|---|
3D Object Tracking | LiDAR Fusion | Object Tracking, MEMC |
Audio-to-text transcription | Audio & Video | TTS |
Basic LiDAR Object Detection | LiDAR Fusion | Object Detection, Autonomous Vehicle |
COCO Object Detection | Image | Common Objects Detection |
Image Sam Embbedding | Image | Image Segmentation |
PointCloud Segment | LiDAR Fusion | Segmentation |
Click on the card to view details.
2. Run the Model
Click the Run Model
button at the top right corner of the details page and select a dataset. You can either predict all classes (or labels) in the model or manually set the confidence level and classes. Then, click Run
.
Model Classes cannot automatically align with Dataset Classes. Therefore, you may need to configure the Model Map before running models.
NOTICE
A team can only run one model at a time.
The "3D Object Tracking" and "Image Sam Embbedding" model can only be run in Annotation Tools.
3. Curate the Results
After running, you can view the results in the Model Runs list. If necessary, click the icon to refresh the list. Click View Dataset
to access the Annotation Tool.
Load results from the model run by selecting the corresponding ID in the top right corner. You can view them on the data and easily compare, analyze, and modify results from different sources.
4. View Model Usage
View your TEAM model usage by clicking View Usage
at the top right corner of the Model Center.
If needed, you can subscribe to upgrade your team plan or directly append model calls :)
Updated 3 months ago