Using the SynthDet example project, you are able to create a custom dataset for your own objects. Creating 3D content for synthetic data and bringing into a Unity project can be challenging. We have provided you some tutorial videos to help with the scanning and import process. The videos below are primarily focused on creating content for object detection ML models. The same approaches can be used for creating digital twins of other objects. For other computer vision tasks there are other strategies that may be employed, including making use of existing assets on the Unity Asset Store or creating assets by hand. For more information, please reach out to us to see how we can help!
| Video link | Description |
|---|---|
| Modeling the Cereal Box | In this video you will see how to model the Quaker Oatmeal squares Box, and this process can be applied for any other box that you would like to model and bring to SynthDet. Following the modeling, you will be shown how to unwrap the UV coordinates of the box for texture placement. |
| Scan Assembly | In this video, you will be shown how to assemble the scanned images of the cereal box into one document, and how to start reconstructing areas that were blurred or muddled in the scans. |
| Matching Scan Edges | In this video, you will be shown how to assemble scans of the cereal box into one complete texture. Additionally, the video will show how to match the edges of the scans to each other and avoid continuity breaks around the product. |
| Exporting the Model and Scanned Texture | In this video, you will be shown how to export the albedo texture for the cereal box from Adobe Photoshop to Unity. Additionally, the video will show how to export the model as an FBX to Unity. |
| Importing into Unity and Configuring Labels in Perception | This video will show you how to bring in an exported model, get a material setup, add a label for it, and add it into Foreground objects in order to use it in Simulations. |