Add model card with attribution

README.md

@@ -1,199 +1,152 @@
 ---
-library_name: transformers
-tags: []
+license: apache-2.0
+base_model: dima806/ai_vs_real_image_detection
+tags:
+- image-classification
+- vision
+- ai-detection
+- deepfake-detection
+- vit
+datasets:
+- CIFAKE
+metrics:
+- accuracy
+- f1
+pipeline_tag: image-classification
 ---
 
-# Model Card for Model ID
+# CapCheck AI Image Detection
 
-<!-- Provide a quick summary of what the model is/does. -->
+Vision Transformer (ViT) fine-tuned for detecting AI-generated images.
 
+## Model Lineage & Attribution
 
+This model builds on the work of others:
 
-## Model Details
+| Layer | Model | Author | License |
+|-------|-------|--------|---------|
+| Base Architecture | [google/vit-base-patch16-224-in21k](https://huggingface.co/google/vit-base-patch16-224-in21k) | Google | Apache 2.0 |
+| AI Detection Fine-tune | [dima806/ai_vs_real_image_detection](https://huggingface.co/dima806/ai_vs_real_image_detection) | dima806 | Apache 2.0 |
+| This Model | capcheck/ai-image-detection | CapCheck | Apache 2.0 |
 
-### Model Description
+**Special thanks to:**
+- **Google** for the Vision Transformer (ViT) architecture
+- **dima806** for fine-tuning on the CIFAKE dataset for AI image detection
 
-<!-- Provide a longer summary of what this model is. -->
+## Model Description
 
-This is the model card of a 🤗 transformers model that has been pushed on the Hub. This model card has been automatically generated.
+- **Architecture**: ViT-Base (86M parameters)
+- **Input Size**: 224x224 pixels
+- **Training Data**: CIFAKE dataset (AI-generated vs real images)
+- **Task**: Binary classification (Real vs Fake/AI-generated)
 
-- **Developed by:** [More Information Needed]
-- **Funded by [optional]:** [More Information Needed]
-- **Shared by [optional]:** [More Information Needed]
-- **Model type:** [More Information Needed]
-- **Language(s) (NLP):** [More Information Needed]
-- **License:** [More Information Needed]
-- **Finetuned from model [optional]:** [More Information Needed]
+## Usage
 
-### Model Sources [optional]
+```python
+from transformers import pipeline
 
-<!-- Provide the basic links for the model. -->
+detector = pipeline("image-classification", model="capcheck/ai-image-detection")
+result = detector("path/to/image.jpg")
 
-- **Repository:** [More Information Needed]
-- **Paper [optional]:** [More Information Needed]
-- **Demo [optional]:** [More Information Needed]
+# Output:
+# [{"label": "Fake", "score": 0.95}, {"label": "Real", "score": 0.05}]
+```
 
-## Uses
+## Labels
 
-<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+| Label | Description |
+|-------|-------------|
+| `Real` | Authentic photograph or real-world image |
+| `Fake` | AI-generated or synthetically created image |
 
-### Direct Use
+## Performance
 
-<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+This model was trained on the CIFAKE dataset. Performance on modern AI generators
+(Flux, Midjourney v6, DALL-E 3, Stable Diffusion 3) may vary.
 
-[More Information Needed]
+See [dima806's model card](https://huggingface.co/dima806/ai_vs_real_image_detection)
+for detailed training metrics.
 
-### Downstream Use [optional]
+## Limitations
 
-<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+- Trained primarily on older AI generators (pre-2024)
+- May have reduced accuracy on:
+  - Very new AI generators not in training data
+  - Heavily compressed images (low JPEG quality)
+  - Images smaller than 224x224 pixels
+- Works best on images with clear subjects
 
-[More Information Needed]
+## Intended Use
 
-### Out-of-Scope Use
+- Content moderation and authenticity verification
+- Research into AI-generated content detection
+- Educational purposes
 
-<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+**Not intended for**:
+- Making consequential decisions without human review
+- Law enforcement evidence without corroborating sources
 
-[More Information Needed]
+## Ethical Considerations
 
-## Bias, Risks, and Limitations
+- This tool is not 100% accurate - false positives harm legitimate creators
+- False negatives can allow misinformation to spread
+- Use in conjunction with other verification methods
+- Human review is recommended for high-stakes decisions
 
-<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+## Roadmap
 
-[More Information Needed]
+### Current Version (v1.0.0)
 
-### Recommendations
+Base model from dima806's CIFAKE-trained ViT. Solid foundation for AI detection.
 
-<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+### Planned Improvements
 
-Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+**Phase 1: Modern Generator Training**
+- Fine-tune on images from Flux, Midjourney v6, DALL-E 3, Stable Diffusion 3
+- Target: Reduce false negatives on 2024+ AI generators
 
-## How to Get Started with the Model
+**Phase 2: False Positive Reduction**
+- Curate dataset of real images commonly flagged as AI
+- Photography edge cases: HDR, heavy editing, digital art
+- Target: <5% false positive rate
 
-Use the code below to get started with the model.
+**Phase 3: Continuous Updates**
+- Quarterly re-training as new generators emerge
+- Community feedback integration
+- Benchmark against latest AI generators
 
-[More Information Needed]
+### Contributing
 
-## Training Details
+We welcome:
+- Dataset contributions (properly licensed images)
+- Bug reports and false positive/negative examples
+- Benchmark results on new generators
 
-### Training Data
+Join the discussion: https://huggingface.co/capcheck/ai-image-detection/discussions
 
-<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+## License
 
-[More Information Needed]
+Apache 2.0 (inherited from Google ViT and dima806's fine-tuned model)
 
-### Training Procedure
+## Citation
 
-<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+If you use this model, please cite:
 
-#### Preprocessing [optional]
+```bibtex
+@misc{capcheck-ai-detection,
+  author = {CapCheck},
+  title = {AI Image Detection Model},
+  year = {2024},
+  publisher = {HuggingFace},
+  url = {https://huggingface.co/capcheck/ai-image-detection},
+  note = {Based on dima806/ai_vs_real_image_detection, fine-tuned from google/vit-base-patch16-224-in21k}
+}
+```
 
-[More Information Needed]
+## Changelog
 
+### v1.0.0 (Initial Release)
 
-#### Training Hyperparameters
-
-- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
-
-#### Speeds, Sizes, Times [optional]
-
-<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
-
-[More Information Needed]
-
-## Evaluation
-
-<!-- This section describes the evaluation protocols and provides the results. -->
-
-### Testing Data, Factors & Metrics
-
-#### Testing Data
-
-<!-- This should link to a Dataset Card if possible. -->
-
-[More Information Needed]
-
-#### Factors
-
-<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
-
-[More Information Needed]
-
-#### Metrics
-
-<!-- These are the evaluation metrics being used, ideally with a description of why. -->
-
-[More Information Needed]
-
-### Results
-
-[More Information Needed]
-
-#### Summary
-
-
-
-## Model Examination [optional]
-
-<!-- Relevant interpretability work for the model goes here -->
-
-[More Information Needed]
-
-## Environmental Impact
-
-<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
-
-Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
-
-- **Hardware Type:** [More Information Needed]
-- **Hours used:** [More Information Needed]
-- **Cloud Provider:** [More Information Needed]
-- **Compute Region:** [More Information Needed]
-- **Carbon Emitted:** [More Information Needed]
-
-## Technical Specifications [optional]
-
-### Model Architecture and Objective
-
-[More Information Needed]
-
-### Compute Infrastructure
-
-[More Information Needed]
-
-#### Hardware
-
-[More Information Needed]
-
-#### Software
-
-[More Information Needed]
-
-## Citation [optional]
-
-<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
-
-**BibTeX:**
-
-[More Information Needed]
-
-**APA:**
-
-[More Information Needed]
-
-## Glossary [optional]
-
-<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
-
-[More Information Needed]
-
-## More Information [optional]
-
-[More Information Needed]
-
-## Model Card Authors [optional]
-
-[More Information Needed]
-
-## Model Card Contact
-
-[More Information Needed]
+- Published base model from dima806/ai_vs_real_image_detection
+- Added proper attribution and documentation
+- Established as CapCheck's source of truth for AI image detection