Hugging Face's logo

OwenElliott
/
image-safety-classifier-s

Image Classification
timm
PyTorch
ONNX
Safetensors
Transformers
Model card Files
xet
 Community
OwenElliott commited on
Commit
015042b
·
verified ·
1 Parent(s): eb8b0b2

Upload README.md with huggingface_hub

Browse files
Files changed (1) hide show
  1. README.md +10 -17
README.md CHANGED
@@ -26,6 +26,16 @@ This model is part of a lineup of image safety classification models in 4 differ
26
  | [`OwenElliott/image-safety-classifier-m`](https://huggingface.co/OwenElliott/image-safety-classifier-m) | 12.1M | 98.06% |
27
  | [`OwenElliott/image-safety-classifier-l`](https://huggingface.co/OwenElliott/image-safety-classifier-l) | 28.5M | 98.20% |
28
 
 
 
 
 
 
 
 
 
 
 
29
  ## Model Usage
30
 
31
  ### Image Classification with timm
@@ -95,23 +105,6 @@ print("Probabilities:", dict(zip(class_names, probs)))
95
  print("Class:", class_names[np.argmax(probs)])
96
  ```
97
 
98
- ## Evaluation
99
-
100
- This model outperforms existing NSFW detectors on our dataset, here we provide an evaluation against [Marqo/nsfw-image-detection-384](https://huggingface.co/Marqo/nsfw-image-detection-384) and [Falconsai/nsfw_image_detection](https://huggingface.co/Falconsai/nsfw_image_detection).
101
-
102
- __NSFW, NSFL, and SFW Evaluation:__
103
-
104
- ![Overall evaluation against other models on NSFW, NSFL and SFW](./images/NSFW-NSFL-SFW-overall-Comparison.png)
105
-
106
- ![Per class evaluation against other models on NSFW, NSFL and SFW](./images/NSFW-NSFL-SFW-overall-per-class-Comparison.png)
107
-
108
- *Note that [Marqo/nsfw-image-detection-384](https://huggingface.co/Marqo/nsfw-image-detection-384) and [Falconsai/nsfw_image_detection](https://huggingface.co/Falconsai/nsfw_image_detection) don't have an explicit NSFL class, for this evaluation the NSFW class from these model is treated as a correct classification for an NSFL image.
109
-
110
- __NSFW and SFW Evaluation:__
111
-
112
- ![Overall evaluation against other models on NSFW and SFW](./images/NSFW-SFW-overall-Comparison.png)
113
- ![Per class accuracy against other models on NSFW and SFW](./images/NSFW-SFW-per-class-Comparison.png)
114
-
115
 
116
  ## Notes on the Dataset
117
 
 
26
  | [`OwenElliott/image-safety-classifier-m`](https://huggingface.co/OwenElliott/image-safety-classifier-m) | 12.1M | 98.06% |
27
  | [`OwenElliott/image-safety-classifier-l`](https://huggingface.co/OwenElliott/image-safety-classifier-l) | 28.5M | 98.20% |
28
 
29
+ ## Evaluation
30
+
31
+ This model outperforms existing NSFW detectors on our dataset, here we provide an evaluation against [Marqo/nsfw-image-detection-384](https://huggingface.co/Marqo/nsfw-image-detection-384) and [Falconsai/nsfw_image_detection](https://huggingface.co/Falconsai/nsfw_image_detection).
32
+
33
+ ![Overall evaluation against other models on NSFW, NSFL and SFW](./images/NSFW-NSFL-SFW-overall-Comparison.png)
34
+
35
+ ![Per class evaluation against other models on NSFW, NSFL and SFW](./images/NSFW-NSFL-SFW-overall-per-class-Comparison.png)
36
+
37
+ *Note that [Marqo/nsfw-image-detection-384](https://huggingface.co/Marqo/nsfw-image-detection-384) and [Falconsai/nsfw_image_detection](https://huggingface.co/Falconsai/nsfw_image_detection) don't have an explicit NSFL class, for this evaluation the NSFW class from these model is treated as a correct classification for an NSFL image.
38
+
39
  ## Model Usage
40
 
41
  ### Image Classification with timm
 
105
  print("Class:", class_names[np.argmax(probs)])
106
  ```
107
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
108
 
109
  ## Notes on the Dataset
110