Cleanup the codebase to make it easier to understand and use

.gitignore

@@ -0,0 +1,12 @@
+__pycache__
+.ipynb_checkpoints
+
+test_imgs
+nsfw_testset
+clip_autokeras_binary_nsfw
+clip_autokeras_nsfw_b32
+TODO
+nsfw-detector.ipynb
+nude-net.ipynb
+test.jpg
+nsfw_trainset

README.md

@@ -1,67 +1,45 @@
 # CLIP-based-NSFW-Detector
 
-This 2 class NSFW-detector is a lightweight Autokeras model that takes CLIP ViT L/14 embbedings as inputs.
-It estimates a value between 0 and 1 (1 = NSFW) and works well with embbedings from images.
-
-DEMO-Colab:
-https://colab.research.google.com/drive/19Acr4grlk5oQws7BHTqNIK-80XGw2u8Z?usp=sharing
-
-The training CLIP V L/14 embbedings can be downloaded here:
-https://drive.google.com/file/d/1yenil0R4GqmTOFQ_GVw__x61ofZ-OBcS/view?usp=sharing (not fully manually annotated so cannot be used as test)
-
-
-The (manually annotated) test set is there https://github.com/LAION-AI/CLIP-based-NSFW-Detector/blob/main/nsfw_testset.zip
-
-https://github.com/rom1504/embedding-reader/blob/main/examples/inference_example.py inference on laion5B
-
-Example of use of the model:
-
-```python
-@lru_cache(maxsize=None)
-def load_safety_model(clip_model):
-    """load the safety model"""
-    import autokeras as ak  # pylint: disable=import-outside-toplevel
-    from tensorflow.keras.models import load_model  # pylint: disable=import-outside-toplevel
-
-    cache_folder = get_cache_folder(clip_model)
-
-    if clip_model == "ViT-L/14":
-        model_dir = cache_folder + "/clip_autokeras_binary_nsfw"
-        dim = 768
-    elif clip_model == "ViT-B/32":
-        model_dir = cache_folder + "/clip_autokeras_nsfw_b32"
-        dim = 512
-    else:
-        raise ValueError("Unknown clip model")
-    if not os.path.exists(model_dir):
-        os.makedirs(cache_folder, exist_ok=True)
-
-        from urllib.request import urlretrieve  # pylint: disable=import-outside-toplevel
-
-        path_to_zip_file = cache_folder + "/clip_autokeras_binary_nsfw.zip"
-        if clip_model == "ViT-L/14":
-            url_model = "https://raw.githubusercontent.com/LAION-AI/CLIP-based-NSFW-Detector/main/clip_autokeras_binary_nsfw.zip"
-        elif clip_model == "ViT-B/32":
-            url_model = (
-                "https://raw.githubusercontent.com/LAION-AI/CLIP-based-NSFW-Detector/main/clip_autokeras_nsfw_b32.zip"
-            )
-        else:
-            raise ValueError("Unknown model {}".format(clip_model))  # pylint: disable=consider-using-f-string
-        urlretrieve(url_model, path_to_zip_file)
-        import zipfile  # pylint: disable=import-outside-toplevel
-
-        with zipfile.ZipFile(path_to_zip_file, "r") as zip_ref:
-            zip_ref.extractall(cache_folder)
-
-    loaded_model = load_model(model_dir, custom_objects=ak.CUSTOM_OBJECTS)
-    loaded_model.predict(np.random.rand(10**3, dim).astype("float32"), batch_size=10**3)
-
-    return loaded_model
-
-
-nsfw_values = safety_model.predict(embeddings, batch_size=embeddings.shape[0])
+The CLIP-based NSFW Detector is a 2-class model primarily trained to detect nudity or pornographic content. It provides an estimation value ranging between 0 and 1, where 1 indicates NSFW content. The detector works well with image embeddings.
+
+Different models are available, ranging from small (`ViT-B-32`) to large models (`ViT-H-14`). Please refer to [models/README.md](models/README.md) for more details.
+
+> **Note**
+> The model files (`clip_autokeras_binary_nsfw.zip, clip_autokeras_nsfw_b32.zip, h14_nsfw.pth, violence_detection_vit_b_32.npy, violence_detection_vit_l_14.npy`) need to stay where they are. Becayse they are used in [clip_retrival](https://github.com/rom1504/clip-retrieval/tree/main) see [link](https://github.com/search?q=repo%3Arom1504%2Fclip-retrieval%20CLIP-based-NSFW-Detector&type=code).
+
+# Local Development
+
+To get started with local development, install the dependencies by running the following command:
+
+```bash
+pip install -r requirements.txt
 ```
 
+# Training
+
+We provide an example for training and testing the `ViT-L-14` openai model, as it's the only model for which we provide the training embeddings. You can find the training embeddings in the [Google Drive link](https://drive.google.com/file/d/1yenil0R4GqmTOFQ_GVw__x61ofZ-OBcS/view?usp=sharing). Please see [data/README.md](data/README.md) for more information.
+
+For training and testing, refer to the notebook [Traininig_Test.ipynb](Traininig_Test.ipynb).
+
+# Inference 
+
+You can find inference examples in the notebook [CLIP_based_NSFW_detector.ipynb](CLIP_based_NSFW_detector.ipynb)
+
+# Additional Resources
+
+Here are some other useful NSFW detectors:
+
+* https://github.com/GantMan/nsfw_model
+* https://github.com/notAI-tech/NudeNet
+
+For NSFW detection datasets, you can refer to:
+
+* https://github.com/alex000kim/nsfw_data_scraper
+* https://archive.org/details/NudeNet_classifier_dataset_v1
+
+
+# LICENSE
+
 This code and model is released under the MIT license:
 
 Copyright 2022, Christoph Schuhmann
@@ -71,3 +49,4 @@ Permission is hereby granted, free of charge, to any person obtaining a copy of
 The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.
 
 THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+