metadata

title: Adult Image Detector
emoji: 🚨
colorFrom: yellow
colorTo: green
sdk: gradio
sdk_version: 4.42.0
app_file: app.py
pinned: false
license: mit

Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference

Adult Image Detector

Model Description

This model is a custom-trained version of YOLOv9-e, pre-trained on a custom dataset. YOLOv9 (You Only Look Once version 9) is a state-of-the-art object detection model known for its speed and accuracy.

Model Details

Model Architecture: YOLOv9-e
Number of Layers: 1,119
Number of Parameters: 69,366,830
GFLOPs: 243.4

Training

The model was trained for 10 epochs on a custom dataset. The training process showed consistent improvement in performance metrics.

Training Hyperparameters

Initial Learning Rate (lr0): 0.070011
Final Learning Rate (lr1, lr2): 0.00208

Training Results

Metric	Initial Value (Epoch 0)	Final Value (Epoch 9)
train/box_loss	1.8995	1.4264
train/cls_loss	2.644	1.1627
train/dfl_loss	1.9846	1.6321
metrics/precision	0.70196	0.69025
metrics/recall	0.44274	0.69178
metrics/mAP_0.5	0.45088	0.7167
metrics/mAP_0.5:0.95	0.27358	0.47964

Performance

The model showed significant improvement over the course of training:

[email protected]: Increased from 0.45088 to 0.7167
[email protected]:0.95: Improved from 0.27358 to 0.47964
Precision: Maintained around 0.69-0.70
Recall: Substantially improved from 0.44274 to 0.69178

Usage

This model can be loaded and used with YOLOv5 compatible frameworks. Here's an example of how to load the model:

from ultralytics import YOLO

model = YOLO('path/to/your/model.pt')
results = model('path/to/image.jpg')

Limitations and Biases

As this model was trained on a custom dataset, it may have biases or limitations specific to that dataset. Users should evaluate the model's performance on their specific use case before deployment.

Additional Information

For more details on the YOLOv9 architecture and its capabilities, please refer to the official YOLOv9 documentation and research paper.