TheKernel01/Tiny-GenImage

1---2dataset_info:3  features:4  - name: image5    dtype: image6  - name: label7    dtype:8      class_label:9        names:10          '0': real11          '1': fake12  - name: generator13    dtype:14      class_label:15        names:16          '0': Real17          '1': ADM18          '2': BigGAN19          '3': GLIDE20          '4': Midjourney21          '5': SD1422          '6': SD1523          '7': VQDM24          '8': Wukong25  splits:26  - name: train27    num_bytes: 655873210328    num_examples: 2800029  - name: validation30    num_bytes: 174876732831    num_examples: 700032  download_size: 835919872333  dataset_size: 830749943134configs:35- config_name: default36  data_files:37  - split: train38    path: data/train-*39  - split: validation40    path: data/validation-*41license: cc-by-nc-sa-4.042task_categories:43- image-classification44language:45- en46---47 48# Tiny GenImage Dataset49 50## 📝 Dataset Description51 52### Dataset Summary53 54The Tiny GenImage Dataset is a curated, scaled-down collection of images and associated metadata designed to train, validate, and benchmark models for detecting and identifying artificially generated content. The dataset contains a mix of real-world images alongside those generated by prominent AI models, including various diffusion models (like Stable Diffusion 1.4/1.5, GLIDE, Midjourney, ADM, VQDM, Wukong) and GANs (BigGAN).55 56Each image is labeled under two categories, enabling researchers and developers to tackle two distinct, high-value computer vision tasks: binary real/fake classification and multi-class source model identification.57 58### Supported Tasks and Leaderboards59 60This dataset directly supports two critical image classification tasks:61 62|**Task ID**|**Task Name**|**Description**|**Output Classes**|63|---|---|---|---|64|**Task A**|Binary Veracity Classification|Classifying images as either real or fake.|2 (real, fake)|65|**Task B**|AI Model Source Identification|Identifying the specific AI generation model used for images labeled as AI-Generated.|9 (Real, ADM, BigGAN, GLIDE, Midjourney, SD14, SD15, VQDM, Wukong)|66 67### Languages68 69The descriptive text, including all class labels and metadata, is in English (en).70 71## 🗂️ Data Splits72 73The dataset is divided into training and validation splits to facilitate standard machine learning workflows.74 75|**Split**|**Number of Instances**|**Notes**|76|---|---|---|77|**train**|28,000|Used for model training and weight optimization.|78|**validation**|7,000|Used for hyperparameter tuning and intermediate model evaluation.|79 80## 💾 Dataset Structure81 82### Data Instances83 84A single data instance consists of an image file and two distinct labels detailing its source and authenticity.85 86|**Field Name**|**Example Value**|**Description**|87|---|---|---|88|**image**|`<PIL.Image.Image object>`|The actual image content loaded into a PIL object.|89|**label**|`1`|Binary label for authenticity (Real vs. AI-Generated).|90|**generator**|`4`|Multi-class label for the specific generation model (or Real).|91 92### Data Fields93 94The dataset contains the following fields:95 96|**Field Name**|**Data Type**|**Description**|97|---|---|---|98|**image**|`datasets.Image()`|The actual image content (e.g., .jpg, .png).|99|**label**|`datasets.ClassLabel`|Task A: Binary label for image veracity.|100|**generator**|`datasets.ClassLabel`|Task B: Label specifying the generation source/model.|101 102## 🏷️ Label Definitions103 104The two label fields use the following strict mappings:105 106**`label` (Binary Veracity Classification)**107 108|**Label**|**Value**|**Description**|109|---|---|---|110|**real**|`0`|Image is a real photograph/non-AI generated.|111|**fake**|`1`|Image was created by an AI generation model.|112 113**`generator` (Model Source Identification)**114 115|**Label**|**Value**|**Description**|116|---|---|---|117|**Real**|`0`|Real image (no AI generation involved).|118|**ADM**|`1`|Generated by Ablated Diffusion Model (Guided Diffusion).|119|**BigGAN**|`2`|Generated by BigGAN.|120|**GLIDE**|`3`|Generated by GLIDE.|121|**Midjourney**|`4`|Generated by Midjourney.|122|**SD14**|`5`|Generated by Stable Diffusion 1.4.|123|**SD15**|`6`|Generated by Stable Diffusion 1.5.|124|**VQDM**|`7`|Generated by Vector Quantized Diffusion Model.|125|**Wukong**|`8`|Generated by the Wukong diffusion model.|126 127## 🔗 Sources128 129- **Original dataset:** [yangsangtai/tiny-genimage (Kaggle)](https://www.kaggle.com/datasets/yangsangtai/tiny-genimage)