| --- |
| tags: |
| - generated_from_trainer |
| model-index: |
| - name: diffusion-detection |
| results: [] |
| license: apache-2.0 |
| datasets: |
| - imagenet-1k |
| metrics: |
| - accuracy |
| pipeline_tag: image-classification |
| --- |
| |
| <!-- This model card has been generated automatically according to the information the Trainer had access to. You |
| should probably proofread and complete it, then remove this comment. --> |
|
|
| # diffusion-detection |
|
|
| This model was trained to distinguish real world images (negative) from machine generated ones (postive). |
|
|
| ## Model usage |
|
|
| ```python |
| from transformers import BeitImageProcessor, BeitForImageClassification |
| from PIL import Image |
| |
| processor = BeitImageProcessor.from_pretrained('TimKond/diffusion-detection') |
| model = BeitForImageClassification.from_pretrained('TimKond/diffusion-detection') |
| |
| image = Image.open("2980_saltshaker.jpg") |
| |
| inputs = processor(images=image, return_tensors="pt") |
| outputs = model(**inputs) |
| logits = outputs.logits |
| |
| predicted_class_idx = logits.argmax(-1).item() |
| print("Predicted class:", model.config.id2label[predicted_class_idx]) |
| ``` |
|
|
| ## Training and evaluation data |
|
|
| [BEiT-base-patch16-224-pt22k](https://huggingface.co/microsoft/beit-base-patch16-224-pt22k) was loaded as a base model for further fine tuning: |
|
|
| As negatives a subsample of 10.000 images from [imagenet-1k](https://huggingface.co/datasets/imagenet-1k) was used. Complementary 10.000 positive images were generated using [Realistic_Vision_V1.4](https://huggingface.co/SG161222/Realistic_Vision_V1.4). |
|
|
| The labels from imagenet-1k were used as prompts for image generation. [GitHub reference](https://github.com/TimKond/diffusion-detection/blob/main/data/DatasetGeneration.py) |
|
|
| ### Training hyperparameters |
|
|
| The following hyperparameters were used during training: |
| - learning_rate: 0.0002 |
| - train_batch_size: 32 |
| - eval_batch_size: 8 |
| - seed: 42 |
| - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08 |
| - lr_scheduler_type: linear |
| - num_epochs: 3 |
| - mixed_precision_training: Native AMP |
|
|
| ### Framework versions |
|
|
| - Transformers 4.29.2 |
| - Pytorch 1.11.0+cu113 |
| - Datasets 2.12.0 |
| - Tokenizers 0.13.3 |
