Initial submission instructions

app.py

@@ -194,6 +194,38 @@ with gr.Blocks(theme=gr.themes.Base()) as demo:
         # # Submit Detector Results Form
         with gr.TabItem("🚀 Submit Detector Results"):
             gr.Markdown("Submit your detector results for evaluation.")
+
+            # Add submission instructions
+            gr.Markdown("""
+            ### Submission Instructions
+            
+            1. Ensure that your detector code follows the [dfd arena repository](https://github.com/BitMind-AI/dfd-arena/tree/main) detectors format. The best way to guarantee compatibility is to develop and test your detector within a local copy of our repo, with dependencies, detector file, and configs in relative locations similar to how we structured our implementations of UCF, NPR, and CAMO detectors.
+
+            **✅ Check list:**
+            - Your code should work with detector dependencies (architecture and training code) imported from a dependencies directory a level above the detector directory.
+              - e.g., `arena/detectors/deepfake_detectors/ucf_detector.py` relies on a dependency folder at `arena/detectors/UCF/`
+              - Our automated benchmarking pipeline will reconstruct the required directory at evaluation time
+            - Implement a `.py` file in `arena/detectors/deepfake_detectors/` containing a `DeepfakeDetector` subclass with PascalCase naming convention, registered as a module in the dfd-arena `DETECTOR_REGISTRY`.
+              - e.g., in `myCustomDetector.py`,
+              ```python
+              @DETECTOR_REGISTRY.register_module(module_name='MyCustomModuleName')
+              class MyCustomDetector(DeepfakeDetector):
+                  # implementation
+              ```
+              - The module name should match the detector name you want to appear on the leaderboard
+            - Create a config YAML file that the DeepfakeDetector loads in `arena/detectors/deepfake_detectors/configs/`.
+            
+            2. Create a HuggingFace model repo with the detector `.py` file, config `.yaml`, and dependencies in the same root level. [Check out our Sample Leaderboard Submission Repo for Reference!](https://huggingface.co/caliangandrew/submit_test/tree/main)
+            
+            3. 📤 Fill out the form below with the correct paths and submit! The results will be processed after a code review by the BitMind team, and an automated test/benchmarking stage.
+            
+            **⚠️ Note:** The Detector Name must match the name of the registered detector module in the dfd arena detector registry. This will be the name of your detector on our leaderboard.
+            - For example, using the [Sample Submission Repo](https://huggingface.co/caliangandrew/submit_test/tree/main) provided, you would submit:  
+              - Detector Name: `test`  
+              - Hugging Face Model Repo: `caliangandrew/submit_test`  
+              - Path to detector `.py`: `test_detector.py`  
+              - Path to config `.YAML`: `test_config.yaml`
+            """)
             
             with gr.Row():
                 detector_name = gr.Textbox(label="Detector Name", placeholder="e.g., MyDetector")