Not enough values to unpack (expected 6, got 1)

Question

I'm loading my object detection model and my classification model to load, detect and predict the class of the image accurately. The code will correctly iterate through the detected objects, filter them based on a confidence threshold of 0.5, and perform classification on each object. If the classification result is "clean" or "dirty," it will print the corresponding message. But the error ValueError: not enough values to unpack (expected 6, got 1) keeps arising again and again

This is my code

#Load an image  to process
import cv2  # Import OpenCV
import numpy as np
from PIL import Image

image_path = '/content/drive/MyDrive/clean_indian2.jpeg'
original_image = cv2.imread(image_path)

#Resize the image to match the expected input size of the YOLOv8 model
resized_image = cv2.resize(original_image, (640, 640)) 

#Convert the resized image to a PyTorch tensor
image_tensor= torch.from_numpy(np.ascontiguousarray(resized_image.transpose(2, 0, 1))).float()

#Normalize the image
image_tensor /= 255.0

#Make it a batch of 1
image_tensor = image_tensor.unsqueeze(0)

#Use YOLOv8n to detect pot seats in the image
#results = yolo_model(image_tensor.unsqueeze(0))  # Add a batch dimension

#Use your YOLOv8 model to predict objects in the image
with torch.no_grad():
    results = yolov8_model(image_tensor)

till here it works fine, but here it gives the error

#Iterate through detected objects and classify them
for obj in results[0]:  # Access the first (and possibly only) prediction
    x1, y1, x2, y2, conf, cls = obj  
#Bounding box coordinates, confidence, and class index

    # You can filter detections based on confidence if needed
    if conf > 0.5:
        object_image = resized_image[int(y1):int(y2), int(x1):int(x2)]

        # Convert the object image to PIL format
        object_pil_image = Image.fromarray(cv2.cvtColor(object_image, cv2.COLOR_BGR2RGB))

        # Perform classification using your classification model
        class_result = classification_model(object_pil_image)

        if class_result == 'clean':
            print("Detected a clean pot seat.")
        elif class_result == 'dirty':
            print("Detected a dirty pot seat.")

Error message:

ValueError:
<ipython-input-74-92bf1423bdc4> in <cell line: 2>()
      1 # Iterate through detected objects and classify them
      2 for obj in results[0]:  # Access the first (and possibly only) prediction
----> 3     x1, y1, x2, y2, conf, cls = obj  # Bounding box coordinates, confidence, and class index
      4 
      5     # You can filter detections based on confidence if needed

ValueError: not enough values to unpack (expected 6, got 1)

Answer 1

Here the obj variable is an ultralytics.engine.results.Results object - a class for storing and manipulating inference results. You cannot unpack it as if you had a list of values. To reach concrete values we need to refer to one of the Results attributes: boxes, masks, probs, keypoints - as the results of detection, segmentation, classification, or pose estimation tasks. Considering you are using the detection task model, we need to refer to the boxes attribute and its properties xyxy, conf, cls - all of them are tensors.

for obj in results[0].boxes:
  x1, y1, x2, y2 = obj.xyxy[0].tolist()
  conf = obj.conf.item()
  cls = int(obj.cls.item())
  print(x1, y1, x2, y2, conf, cls)

The estimated output format will be the following:

2643.072265625 1688.5892333984375 3643.5322265625 2735.485107421875 0.4271290600299835 77

More about working with yolov8 prediction results is in the documentation: https://docs.ultralytics.com/modes/predict/#working-with-results