Inconsistent results while using adr/6 model as an API and in browser

Hi,
Currently I am using adr/6 model for object detection on images. while using the same image within my application and invoking model using api call with “prediction = self.client.infer(img, model_id=self.model_id)” response show 0 detections. whereas If I use same image in browser it shows 3 detections. I tried debugging but visualising if image input if it’s causing any issue but no problems there. Why are results inconsistent I am unable to debug. Plz help I need to solve this as soon as possible.

current inference code:
# Create debug directory if it doesn’t exist
debug_dir = “Backend/debug_visualizations/roboflow_input”
os.makedirs(debug_dir, exist_ok=True)

        # Convert bytes to base64 string
        # base64_image_string = self._bytes_to_base64_string(image_bytes)
        nparr = np.frombuffer(image_bytes, np.uint8)
        img = cv2.imdecode(nparr, cv2.IMREAD_COLOR)
        if img is None:
            raise ValueError("Failed to decode image")
        # Perform inference
            # Save the input image with timestamp for debugging
        timestamp = datetime.now().strftime("%Y%m%d_%H%M%S")
        debug_path = os.path.join(debug_dir, f"roboflow_input_{timestamp}.png")
        cv2.imwrite(debug_path, img)
        logger.info(f"Saved debug input image to: {debug_path}")
        prediction = self.client.infer(img, model_id=self.model_id)

        boxes_xyxy = []
        classes = []
        scores = [x]

        # Ensure prediction is a dictionary and has 'predictions' key
        if not isinstance(prediction, dict):
             logger.error(f"Unexpected prediction format: {type(prediction)}. Expected dict.")
             prediction = {} # Avoid further errors

        predictions_data = prediction.get('predictions', [])
        num_detections = len(predictions_data)
        logger.info(f"Prediction received: {prediction}")
        logger.info(f"Received {num_detections} predictions.")
        logger.info(f"Predictions data: {predictions_data}") 

and logs output from this code is
INFO - Prediction received: {‘inference_id’: ‘12898b8a-d898-4d21-89f3-5f6e6aa76b2e’, ‘time’: 0.037813304999872344, ‘image’: {‘width’: 800, ‘height’: 1095}, ‘predictions’: }
INFO - Received 0 predictions.
INFO - Predictions data:
INFO - Number of detections: 0
INFO - Yolo/Roboflow Inference: boxes → , classes → , scores → , num_detections → 0

This topic was automatically closed 21 days after the last reply. New replies are no longer allowed.