Datasets:
Update README.md
Browse files
README.md
CHANGED
@@ -47,7 +47,7 @@ large dataset of SCC brightfield images.
|
|
47 |
The results of these models were pretty impressive, but still needed many images.
|
48 |
Hence, this dataset was created to test the capabilities of dreambooth on brightfield microscopy image generation.
|
49 |
I'm testing several configurations:
|
50 |
-
- Diffusion Model Architectures (SD-1.5, SD-2.1, SDXL 1.0)
|
51 |
- Training Data Size (10, 20, 30, 50)
|
52 |
- 4 Concepts are trained in parallel (cell, cell rug, well edge, debris)
|
53 |
- With and without subject class images for class-specific prior preservation loss impact assessment
|
@@ -59,7 +59,7 @@ The dataset consists of several classes:
|
|
59 |
- Generated images from SDXL 1.0, one class for each concept
|
60 |
|
61 |
These classes are used in the concepts for the dreambooth model training,
|
62 |
-
resulting in
|
63 |
Unfortunately, due to time constraints, I'm not able to test many hyperparameter configurations for each model, nor play around
|
64 |
a lot with prompt engineering.
|
65 |
-
This research serves as a base others (or me) can work upon.
|
|
|
47 |
The results of these models were pretty impressive, but still needed many images.
|
48 |
Hence, this dataset was created to test the capabilities of dreambooth on brightfield microscopy image generation.
|
49 |
I'm testing several configurations:
|
50 |
+
- Diffusion Model Architectures (SD-1.5(, SD-2.1, SDXL 1.0)) -- The last two had to be discontinued due to time, and compute constraints
|
51 |
- Training Data Size (10, 20, 30, 50)
|
52 |
- 4 Concepts are trained in parallel (cell, cell rug, well edge, debris)
|
53 |
- With and without subject class images for class-specific prior preservation loss impact assessment
|
|
|
59 |
- Generated images from SDXL 1.0, one class for each concept
|
60 |
|
61 |
These classes are used in the concepts for the dreambooth model training,
|
62 |
+
resulting in 8 models trained to assess the usability of dreambooth in this domain.
|
63 |
Unfortunately, due to time constraints, I'm not able to test many hyperparameter configurations for each model, nor play around
|
64 |
a lot with prompt engineering.
|
65 |
+
This research serves as a base thath others (or me) can work upon.
|