Add /api/detect endpoint for JSON detections (MonitaQC compatible)

app.py

@@ -269,7 +269,22 @@ with gr.Blocks(title="Industrial Defect Detection") as demo:
         fn=gradio_inference,
         inputs=[input_image, model_dropdown, conf_slider],
         outputs=output_image,
-        api_name="predict"  # Creates /api/predict endpoint
+        api_name="predict"  # Creates /api/predict endpoint (returns image)
+    )
+
+    # Hidden interface for JSON API (for MonitaQC compatibility)
+    with gr.Row(visible=False):
+        json_image = gr.Image(type="numpy")
+        json_model = gr.Dropdown(choices=[v["name"] for v in MODELS.values()])
+        json_conf = gr.Slider(minimum=0.0, maximum=1.0, value=0.25)
+        json_output = gr.JSON()
+        json_btn = gr.Button("JSON Detect")
+
+    json_btn.click(
+        fn=api_inference,
+        inputs=[json_image, json_model, json_conf],
+        outputs=json_output,
+        api_name="detect"  # Creates /api/detect endpoint (returns JSON)
     )
 
     gr.Markdown("### Example Images")
@@ -285,16 +300,17 @@ with gr.Blocks(title="Industrial Defect Detection") as demo:
     gr.Markdown("""
     ### API Access
 
-    This Space provides an API endpoint that can be accessed programmatically:
+    This Space provides two API endpoints:
 
-    - **API Endpoint**: `/api/predict`
-    - **Method**: POST
-    - **Parameters**:
-      - `data[0]`: image (base64 or file)
-      - `data[1]`: model name (e.g., "Data Matrix")
-      - `data[2]`: confidence threshold (0.0-1.0)
+    **1. Image API** (returns annotated image):
+    - **Endpoint**: `/api/predict`
+    - **Returns**: Annotated image with bounding boxes
 
-    **Python Example:**
+    **2. JSON API** (returns detection data - for MonitaQC):
+    - **Endpoint**: `/api/detect`
+    - **Returns**: JSON array of detections with bboxes and confidence
+
+    **Python Example (Image):**
     ```python
     from gradio_client import Client
 
@@ -307,7 +323,21 @@ with gr.Blocks(title="Industrial Defect Detection") as demo:
     )
     ```
 
-    **Model Names:**
+    **Python Example (JSON - for MonitaQC):**
+    ```python
+    from gradio_client import Client
+
+    client = Client("smartfalcon-ai/Industrial-Defect-Detection")
+    detections = client.predict(
+        "path/to/image.jpg",
+        "Data Matrix",
+        0.25,
+        api_name="/detect"
+    )
+    # Returns: [{"bbox": [x1, y1, x2, y2], "confidence": 0.85, "class_id": 0, ...}]
+    ```
+
+    **Available Models:**
     - Data Matrix
     - Dental Implant
     - Ball Pen