Create README.md

README.md

@@ -0,0 +1,37 @@
+---
+datasets:
+- Hemg/AI-Generated-vs-Real-Images-Datasets
+metrics:
+- accuracy
+base_model:
+- microsoft/resnet-50
+pipeline_tag: image-classification
+tags:
+- art
+---
+
+# Multi-Task Image Classifier
+
+
+## Model Overview
+This model is a **Multi-Task Image Classifier** that performs two tasks simultaneously:
+1. **Object Recognition:** Identifies the primary object in an image (e.g., "cat," "dog," "car," etc.).
+2. **Authenticity Classification:** Determines whether the image is AI-generated or a real photograph.
+
+The model uses a ResNet-50 backbone with two heads: one for multi-class object recognition (trained on pseudo-labels generated by a Vision Transformer) and another for binary classification (AI-generated vs. Real). It was trained on a subset of the [Hemg/AI-Generated-vs-Real-Images-Datasets](https://huggingface.co/datasets/Hemg/AI-Generated-vs-Real-Images-Datasets).
+
+## Intended Use
+This model is designed for:
+- **Digital Content Verification:** Detecting AI-generated images to prevent misinformation.
+- **Social Media Moderation:** Automatically flagging images that are likely AI-generated.
+- **Content Analysis:** Helping researchers understand the prevalence of AI art versus real images.
+
+## How to Use
+You can use this model locally or via a Hugging Face Space. For local usage, load the state dictionary into the model architecture using PyTorch. For example:
+```python
+import torch
+from model import MultiTaskModel  # your model definition
+
+model = MultiTaskModel(...)
+model.load_state_dict(torch.load("multitask_model_weights.pth", map_location="cpu"))
+model.eval()