Erax NSFW V1.0

🚀 EraX-NSFW-V1.0

A Highly Efficient Model for NSFW Detection, effective for pre - publication image and video control, and limiting children's access to harmful publications.

⚠️ Important Note

THIS MEDIA CONTAINS SENSITIVE CONTENT (I.E. NUDITY, VIOLENCE, PROFANITY, PORN) THAT SOME PEOPLE MAY FIND OFFENSIVE. YOU MUST BE 18 OR OLDER TO VIEW THIS CONTENT.

🚀 Quick Start

EraX-NSFW-V1.0 is a powerful model for NSFW detection. You can either just predict the classes and their bounding boxes or even mask the predicted harmful object(s) or mask the entire image. Please see the deployment codes below.

Installation

pip install ultralytics supervision huggingface-hub

Download Pretrained model

from huggingface_hub import snapshot_download
snapshot_download(repo_id="erax-ai/EraX-NSFW-V1.0", local_dir="./", force_download=True)

Simple Use Case

from ultralytics import YOLO
from PIL import Image
import supervision as sv
import numpy as np

IOU_THRESHOLD        = 0.3
CONFIDENCE_THRESHOLD = 0.2

# pretrained_path = "erax_nsfw_yolo11n.pt"
# pretrained_path = "erax_nsfw_yolo11s.pt"
pretrained_path = "erax_nsfw_yolo11m.pt"

image_path_list = ["test_images/img_1.jpg", "test_images/img_2.jpg"]

model = YOLO(pretrained_path)
results = model(image_path_list,
                  conf=CONFIDENCE_THRESHOLD,
                  iou=IOU_THRESHOLD
                )


for result in results:
    annotated_image = result.orig_img.copy()
    h, w = annotated_image.shape[:2]
    anchor = h if h > w else w

    detections = sv.Detections.from_ultralytics(result)
    label_annotator = sv.LabelAnnotator(text_color=sv.Color.BLACK,
                                        text_position=sv.Position.CENTER,
                                        text_scale=anchor/1700)
    
    pixelate_annotator = sv.PixelateAnnotator(pixel_size=anchor/50)
    
    annotated_image = pixelate_annotator.annotate(
        scene=annotated_image.copy(),
        detections=detections
    )


    annotated_image = label_annotator.annotate(
        annotated_image,
        detections=detections
    )

    
    sv.plot_image(annotated_image, size=(10, 10))

✨ Features

• High Efficiency: Effectively used for pre - publication image and video control, and limiting children's access to harmful publications.
• Flexible Usage: You can predict classes and bounding boxes, or mask harmful objects or the entire image.

📦 Installation

To use the trained model, follow these steps:

1. Install the necessary packages:

pip install ultralytics supervision huggingface-hub

2. Download Pretrained model:

from huggingface_hub import snapshot_download
snapshot_download(repo_id="erax-ai/EraX-NSFW-V1.0", local_dir="./", force_download=True)

💻 Usage Examples

Basic Usage

from ultralytics import YOLO
from PIL import Image
import supervision as sv
import numpy as np

IOU_THRESHOLD        = 0.3
CONFIDENCE_THRESHOLD = 0.2

# pretrained_path = "erax_nsfw_yolo11n.pt"
# pretrained_path = "erax_nsfw_yolo11s.pt"
pretrained_path = "erax_nsfw_yolo11m.pt"

image_path_list = ["test_images/img_1.jpg", "test_images/img_2.jpg"]

model = YOLO(pretrained_path)
results = model(image_path_list,
                  conf=CONFIDENCE_THRESHOLD,
                  iou=IOU_THRESHOLD
                )


for result in results:
    annotated_image = result.orig_img.copy()
    h, w = annotated_image.shape[:2]
    anchor = h if h > w else w

    detections = sv.Detections.from_ultralytics(result)
    label_annotator = sv.LabelAnnotator(text_color=sv.Color.BLACK,
                                        text_position=sv.Position.CENTER,
                                        text_scale=anchor/1700)
    
    pixelate_annotator = sv.PixelateAnnotator(pixel_size=anchor/50)
    
    annotated_image = pixelate_annotator.annotate(
        scene=annotated_image.copy(),
        detections=detections
    )


    annotated_image = label_annotator.annotate(
        annotated_image,
        detections=detections
    )

    
    sv.plot_image(annotated_image, size=(10, 10))

📚 Documentation

Model Details / Overview

Labels

Training Configuration

• Model Weights Files:
  • Nano: erax_nsfw_yolo11n.pt
  • Small: erax_nsfw_yolo11s.pt
  • Medium: erax_nsfw_yolo11m.pt
• Number of Epochs: 100
• Learning Rate: 0.01
• Batch Size: 208
• Image Size: 640x640
• Training server: 8 x NVIDIA RTX A4000 (16GB GDDR6)
• Training time: ~10 hours

Training Validation Results

Training and Validation Losses

Confusion Matrix

More examples

1. Example 01:
2. Example 02:
3. Example 03: SAFEEST for using make_love class as it will cover entire context. Without make_love class | With make_love class :-------------------------:|:-------------------------: | | |

🔧 Technical Details

Inference

To use the trained model, follow these steps:

1. Install the necessary packages.
2. Download the pretrained model.
3. Use the simple use case code to perform inference.

Training

Scripts for training: https://github.com/EraX-JS-Company/EraX-NSFW-V1.0

📄 License

This project is licensed under the Apache 2.0 License.

Citation

If you find our project useful, we would appreciate it if you could star our repository and cite our work as follows:

@article{EraX-NSFW-V1.0,
  author    = {Phạm Đình Thục and
              Mr. Nguyễn Anh Nguyên and
              Đoàn Thành Khang and
              Mr. Trần Hải Khương and
              Mr. Trương Công Đức and 
              Phan Nguyễn Tuấn Kha and 
              Phạm Huỳnh Nhật},
  title     = {EraX-NSFW-V1.0: A Highly Efficient Model for NSFW Detection},
  organization={EraX JS Company},
  year={2024},
  url={https://huggingface.co/erax-ai/EraX-NSFW-V1.0}
}