Update README.md
Browse files
README.md
CHANGED
@@ -25,3 +25,27 @@ File lists:
|
|
25 |
|
26 |
These model is used for monochrome image classification, based on CNNs and Transformers, trained with dataset [deepghs/monochrome_danbooru(private)](https://huggingface.co/datasets/deepghs/monochrome_danbooru).
|
27 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
25 |
|
26 |
These model is used for monochrome image classification, based on CNNs and Transformers, trained with dataset [deepghs/monochrome_danbooru(private)](https://huggingface.co/datasets/deepghs/monochrome_danbooru).
|
27 |
|
28 |
+
The following are the checkpoints that have been formally put into use, all based on the Caformer architecture:
|
29 |
+
|
30 |
+
| Checkpoint | Algorithm | Accuracy | False Negative | False Positive |
|
31 |
+
|:----------------------------:|:---------:|:----------:|:--------------:|:--------------:|
|
32 |
+
| monochrome-caformer-40 | caformer | 96.41% | 2.69% | 0.89% |
|
33 |
+
| **monochrome-caformer-110** | caformer | **96.97%** | 1.57% | 1.46% |
|
34 |
+
| monochrome-caformer_safe2-80 | caformer | 94.84% | **1.12%** | 4.03% |
|
35 |
+
| monochrome-caformer_safe4-70 | caformer | 94.28% | **0.67%** | 5.04% |
|
36 |
+
|
37 |
+
**`monochrome-caformer-110` has the best overall accuracy** among them, but considering that this model is often used to screen out monochrome images
|
38 |
+
and we want to screen out as many as possible without omission, we have also introduced weighted models (`safe2` and `safe4`).
|
39 |
+
Although their overall accuracy has been slightly reduced, the probability of False Negative (misidentifying a monochrome image as a colored one) is lower,
|
40 |
+
making them more suitable for batch screening.
|
41 |
+
|
42 |
+
## Deepdanbooru
|
43 |
+
|
44 |
+
`deepdanbooru` is a model used to tag anime images. Here, we provide a table for tag classification called `deepdanbooru_tags.csv`,
|
45 |
+
as well as an ONNX model (from [chinoll/deepdanbooru](https://huggingface.co/spaces/SmilingWolf/wd-v1-4-tags)).
|
46 |
+
|
47 |
+
It's worth noting that due to the poor quality of the deepdanbooru model itself and the relatively old dataset,
|
48 |
+
it is only for testing purposes and is not recommended to be used as the main classification model. We recommend using the `wd14` model instead, see:
|
49 |
+
|
50 |
+
* https://huggingface.co/spaces/SmilingWolf/wd-v1-4-tags
|
51 |
+
|