fix structure

app/app.py → app.py

@@ -63,11 +63,8 @@ model = CLIPImageClassifier(
 ).to(device)
 
 # Load Weights
-weights_path = os.path.join(
-    os.path.dirname(os.path.dirname(os.path.abspath(__file__))),
-    "model",
-    "best_clip_ai_detector.pth",
-)
+weights_path = "best_clip_ai_detector.pth",
+
 print(f"Loading weights from {weights_path}...")
 try:
     model.load_state_dict(torch.load(weights_path, map_location=device))

app/README.md

@@ -1,24 +0,0 @@
-# AI Image Detector Application
-
-This directory contains the deployment code for the AI Image Detector. It provides a simple web interface using Gradio, allowing users to upload images and receive real-time predictions on whether the image is real or AI-generated.
-
-## Files
-
-- `app.py`: The main Python script that loads the trained model and launches the Gradio interface.
-
-## Setup and Usage
-
-1. Ensure you have installed the project dependencies listed in the root `requirements.txt`.
-2. Ensure the trained model weights file `best_clip_ai_detector.pth` is located in the `../model/` directory relative to this folder.
-3. Run the application:
-
-   ```bash
-   python app.py
-   ```
-
-4. The script will initialize the model (downloading the base CLIP model if necessary) and start a local web server.
-5. Access the interface via the URL displayed in the console (typically `http://127.0.0.1:7860`).
-
-## Model Details
-
-The application uses a `CLIPImageClassifier` class that loads a pre-trained `openai/clip-vit-base-patch32` model from the `transformers` library. It adds a custom classification head on top of the CLIP vision model to perform binary classification (Real vs. AI-Generated).