Data
Toy Dataset
To test vodex we are using a simple toy dataset, where each image frame is labeled with the experimental conditions as well as the frame number and volume slice.
You can download the toy dataset from GitHub.
Image Data structure
The toy dataset movie consists of 42 frames.
Each 10 frames make one volume, so the whole recording consists of 4 full volumes and two extra frames at the end of the recording.
The movie is split into 3 TIFF files:
- 7 frames in the first file,
- 18 frames in the second file and
- 17 frames in the third one.
Time Annotation Structure
We are tracking 3 types of toy experimental conditions in this dataset:
1. c label: a condition label that appears in the middle of the screen. It takes c1, c2, c3 values.
c label is a Cycle: | The movie starts with a c1 label for 10 frames, then c2 for 10 frames, followed by c3 for another 10 frames. This pattern repeats until the end of the recording |
---|---|
c1 | 10 frames |
c2 | 10 frames |
c3 | 10 frames |
2. shape: a circle or a square shapes on the screen.
shape is a Cycle: | There is circle on the screen for the first 5 frames, then a square for 10 frames, then a circle again for 5 frames. This pattern repeats until the end of the recording. |
---|---|
c | 5 frames |
s | 10 frames |
c | 5 frames |
3. light: whether the background is bright or dark, with the light being on or off respectively.
light is a Timeline: | The light is off for the first 10 frames, then on for 20 frames and then off again for 12 frames. |
---|---|
off | 10 frames |
on | 20 frames |
off | 12 frames |
See a frame-by-frame labels in the image on the left [click on image to zoom in].
Database
You can find an example of the database for the Toy Dataset on GitHub.