Demos
READY demo aplication (ready.py) is running in a local host LaptopGPU with NVIDIARTXA2000-8GB using local-built holoscan-sdk.
UNet was trained in cricket with A100-40GB and using 27.4K images of 1 channel.
| Animation | Data, Model(s), API |
|---|---|
Fig. Initial demo using UNET with OpenEDS datsets. Video has 3 frames copied 10 times to create a 30 frame per second video. |
DATA: OpenEDS; MODEL: UNET; API: ready.py d98c497 |
Fig. Pupil segmentation was masked to compute its centroid values as coordinates. Centroid values for x are plotted as cummulative time series of 30 samples. Streamed video consist of different 6 frames copied 5 times to create a 30 frames per second video. |
DATA: OpenEDS; MODEL: UNET; API: ready.py b44515a |
Fig. Initial demo using UNET with OpenEDS datsets. Video has 3 frames copied 10 times to create a 30 frame per second video.
Fig. Pupil segmentation was masked to compute its centroid values as coordinates. Centroid values for x are plotted as cummulative time series of 30 samples. Streamed video consist of different 6 frames copied 5 times to create a 30 frames per second video.