## Dataset Information
The dataset focuses on the printed part of 316ss track prints that were one a 316ss substrate. This was done in-situ and using Lawrence Livermore National Laboratory on board High Speed Camera video recorder for laser powder bed fusion.
This dataset includes 200 videos that view the whole printing process from starting to print to finishing.
- Raw images and corresponding groud truth masks are [here][1].
**Metadata:**
- Maker = Mikrotron
- Model = MC1362
- Detector = CMOS Sensor
- dimension size = 1280x1024
- Capture Rate = 1kHz
- wavelength detection = 1024 gray levels
- resolution limit = 14 μm/pixel
- Orientation = Co-axial
- filter band pass = 780−820 nm
- video output dimensions = 256 × 256
- measure scale = 17 μm/pixel
- exposure time = 200 μs
**Want to contribute? Fill up this [Google Form][2]**.
<h2>
Benchmarks
</h2>
| Model | Description | Binary IoU | Library | Input Image Size | Performance Metrics | Paper | Evaluation Script |
|:---------------------------------:|:--------------------------------------------:|:----------:|:--------------------:|:----------------:|:------------------------------------------:|:------------------------------------------:|:------------------------------------------:|
| [unet.h5][3] | U-Net model with 4-block encoder and decoder | -|Tensorflow/Keras | 1024 x 1024| Accuracy <br> Precision <br> Recall <br> Binary IoU* | - | - |
[1]: https://osf.io/dzv3b/files/osfstorage
[2]: https://forms.gle/4pGRhkdw4wGNy1dY9
[3]: https://osf.io/download/6658e77e65e1de4893894165/
