initial

app.py

@@ -3,17 +3,21 @@ from inference_sdk import InferenceHTTPClient
 from PIL import Image, ImageDraw, ImageEnhance
 import matplotlib.pyplot as plt
 import os
+import tempfile
+import io
 from collections import defaultdict
+import numpy as np
 
+# Initialize the Roboflow client with API key from environment variable
 API_KEY = os.getenv("ROBOFLOW_API_KEY")
-
-# Initialize the Roboflow client
 CLIENT = InferenceHTTPClient(
     api_url="https://detect.roboflow.com",
     api_key=API_KEY
 )
 
-MODEL_ID = "hvac-trial/1"
+# Model ID is predefined
+MODEL_ID = "hvac-trial/1"  # Replace with your actual model ID if different
+
 # Define colors for each component type
 COLORS = {
     "BYPASS DAMPER": (255, 0, 255),  # Magenta
@@ -35,7 +39,7 @@ def enhance_image(image):
     image = ImageEnhance.Contrast(image).enhance(1.2)
     return image.convert('RGB')  # Convert back to RGB
 
-def detect_components(image_pil, confidence_threshold, grid_size, model_id):
+def detect_components(image_pil, confidence_threshold, grid_size):
     """
     Detects components in an image using the AI model with segmentation.
     
@@ -43,10 +47,9 @@ def detect_components(image_pil, confidence_threshold, grid_size, model_id):
         image_pil: PIL Image object
         confidence_threshold: Minimum confidence level for detections
         grid_size: Tuple (rows, cols) for grid segmentation
-        model_id: Roboflow model ID
     
     Returns:
-        Tuple of (final image with detections, component counts)
+        Tuple of (final image with detections, component counts, pass results)
     """
     width, height = image_pil.size
     grid_rows, grid_cols = grid_size
@@ -86,7 +89,7 @@ def detect_components(image_pil, confidence_threshold, grid_size, model_id):
                 
                 try:
                     # Run inference
-                    result = CLIENT.infer(segment_path, model_id=model_id)
+                    result = CLIENT.infer(segment_path, model_id=MODEL_ID)
                     filtered_predictions = [pred for pred in result["predictions"] 
                                            if pred["confidence"] >= confidence_threshold]
                     
@@ -143,11 +146,11 @@ def detect_components(image_pil, confidence_threshold, grid_size, model_id):
         )
     
     # Create a table of counts
-    count_table = "\n".join([f"{label}: {count}" for label, count in final_counts.items()])
+    count_table = "\n".join([f"{label}: {count}" for label, count in sorted(final_counts.items())])
     
     return final_image, count_table, pass_results
 
-def process_image(input_image, confidence_threshold, grid_rows, grid_cols, model_id):
+def process_image(input_image, confidence_threshold, grid_rows, grid_cols):
     """
     Process the input image and return results for Gradio interface.
     
@@ -155,14 +158,10 @@ def process_image(input_image, confidence_threshold, grid_rows, grid_cols, model
         input_image: Input image (numpy array from Gradio)
         confidence_threshold: Confidence threshold slider value
         grid_rows, grid_cols: Grid size
-        model_id: Model ID input
     
     Returns:
         Tuple of (final image, component counts, pass images)
     """
-    if model_id.strip() == "":
-        return None, "Error: Please provide a valid Model ID", None, None, None
-    
     if API_KEY is None:
         return None, "Error: ROBOFLOW_API_KEY environment variable is not set", None, None, None
     
@@ -174,8 +173,7 @@ def process_image(input_image, confidence_threshold, grid_rows, grid_cols, model
         final_image, count_table, pass_results = detect_components(
             input_pil, 
             confidence_threshold, 
-            (grid_rows, grid_cols),
-            model_id
+            (grid_rows, grid_cols)
         )
         
         # Convert individual pass images for display
@@ -200,7 +198,6 @@ with gr.Blocks(title="HVAC Component Detection") as demo:
     with gr.Row():
         with gr.Column():
             input_image = gr.Image(label="Input Image", type="numpy")
-            model_id = gr.Textbox(label="Roboflow Model ID", placeholder="Enter your model ID")
             confidence = gr.Slider(minimum=0.1, maximum=1.0, value=0.3, step=0.05, 
                                 label="Confidence Threshold")
             
@@ -221,17 +218,16 @@ with gr.Blocks(title="HVAC Component Detection") as demo:
     
     submit_btn.click(
         fn=process_image,
-        inputs=[input_image, confidence, grid_rows, grid_cols, model_id],
+        inputs=[input_image, confidence, grid_rows, grid_cols],
         outputs=[output_image, component_counts, pass1_output, pass2_output, pass3_output]
     )
     
     gr.Markdown("""
     ## Instructions
-    1. Enter your Roboflow Model ID in the text field
-    2. Upload an image containing HVAC components
-    3. Adjust the confidence threshold and grid size if needed
-    4. Click "Detect Components" to start the detection process
-    5. View the results from all three detection passes
+    1. Upload an image containing HVAC components
+    2. Adjust the confidence threshold and grid size if needed
+    3. Click "Detect Components" to start the detection process
+    4. View the results from all three detection passes
     
     Note: This application requires a ROBOFLOW_API_KEY environment variable.
     """)

requirements.txt

@@ -1,4 +1,5 @@
 inference-sdk
 pillow
 matplotlib
-gradio
+gradio
+numpy