predicting from 2D array

[polinamalova0]

Search before asking

• I have searched the YOLOv5 issues and discussions and found no similar questions.

Question

Thanks YOLO team for making such a great tool! I have question regarding the prediction of the already trained model.
what do I have as an input: an np.array with the shape (428, 428). each value of the array varies from 1.3 to 1.4
I have faced the problem that when I am trying to predict, the model says that there must be specific data format given in the form of (n, 3, w, h). Therefore, I transposed the array and provided it in the correct form to the model.

    print('input image shape: ', input_data.shape)  # (428, 428)
    norm_input_data = (input_data- np.min(input_data)) / (np.max(input_data) - np.min(input_data))
    norm_rgb_input_data = cv2.cvtColor((norm_input_data * 255).astype(np.uint8), cv2.COLOR_GRAY2RGB)
    rgb_images = [norm_rgb_input_data]

    cvImage_array = np.array(rgb_images).transpose((0, 3, 1, 2))
    print('cvImage_array.shape', cvImage_array.shape) # (1, 3, 428, 428)
    print('cvImage_array.dtype', cvImage_array.dtype) # uint8
    results = model(cvImage_array)

however, in the end i get this error, and i have no clue how to proceed, since all the dimentions are correct and when i print out cvImage_array i do get the values in the end.

Traceback (most recent call last):
  File "path\max_proj.py", line 74, in <module>
    results = model(cvImage_array)
              ^^^^^^^^^^^^^^^^^^^^
  File "path\model.py", line 176, in __call__
    return self.predict(source, stream, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "path\model.py", line 451, in predict
    return self.predictor.predict_cli(source=source) if is_cli else self.predictor(source=source, stream=stream)
                                                                    ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "path\predictor.py", line 168, in __call__
    return list(self.stream_inference(source, model, *args, **kwargs))  # merge list of Result into one
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "path\_contextlib.py", line 35, in generator_context
    response = gen.send(None)
               ^^^^^^^^^^^^^^
  File "path\predictor.py", line 244, in stream_inference
    im = self.preprocess(im0s)
         ^^^^^^^^^^^^^^^^^^^^^
  File "path\predictor.py", line 124, in preprocess
    im = np.stack(self.pre_transform(im))
                  ^^^^^^^^^^^^^^^^^^^^^^
  File "path\predictor.py", line 156, in pre_transform
    return [letterbox(image=x) for x in im]
            ^^^^^^^^^^^^^^^^^^
  File "path\augment.py", line 771, in __call__
    img = cv2.resize(img, new_unpad, interpolation=cv2.INTER_LINEAR)
          ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
cv2.error: OpenCV(4.9.0) D:\a\opencv-python\opencv-python\opencv\modules\imgproc\src\resize.cpp:3789: error: (-215:Assertion failed) !dsize.empty() in function 'cv::hal::resize'

I would be very grateful for the help. Thank you in advance!

Additional

No response

[github-actions]

👋 Hello @polinamalova0, thank you for your interest in YOLOv5 🚀! Please visit our ⭐️ Tutorials to get started, where you can find quickstart guides for simple tasks like Custom Data Training all the way to advanced concepts like Hyperparameter Evolution.

If this is a 🐛 Bug Report, please provide a minimum reproducible example to help us debug it.

If this is a custom training ❓ Question, please provide as much information as possible, including dataset image examples and training logs, and verify you are following our Tips for Best Training Results.

Requirements

Python>=3.8.0 with all requirements.txt installed including PyTorch>=1.8. To get started:

git clone https://github.com/ultralytics/yolov5  # clone
cd yolov5
pip install -r requirements.txt  # install

Environments

YOLOv5 may be run in any of the following up-to-date verified environments (with all dependencies including CUDA/CUDNN, Python and PyTorch preinstalled):

• Notebooks with free GPU:
• Google Cloud Deep Learning VM. See GCP Quickstart Guide
• Amazon Deep Learning AMI. See AWS Quickstart Guide
• Docker Image. See Docker Quickstart Guide

Status

If this badge is green, all YOLOv5 GitHub Actions Continuous Integration (CI) tests are currently passing. CI tests verify correct operation of YOLOv5 training, validation, inference, export and benchmarks on macOS, Windows, and Ubuntu every 24 hours and on every commit.

Introducing YOLOv8 🚀

We're excited to announce the launch of our latest state-of-the-art (SOTA) object detection model for 2023 - YOLOv8 🚀!

Designed to be fast, accurate, and easy to use, YOLOv8 is an ideal choice for a wide range of object detection, image segmentation and image classification tasks. With YOLOv8, you'll be able to quickly and accurately detect objects in real-time, streamline your workflows, and achieve new levels of accuracy in your projects.

Check out our YOLOv8 Docs for details and get started with:

pip install ultralytics

[glenn-jocher]

Hello! 👋 It's great to see your enthusiasm for working with YOLOv5. The error you're encountering often points to an issue with the input dimensions or type at some point before it's passed into the model. Your preprocessing steps seem correct, converting the grayscale image to RGB and adjusting its shape to match the expected input format of (n, 3, w, h).

Given the error trace, it appears the issue might not be with your code directly but rather an unexpected interaction with OpenCV or the way the processed image array is being interpreted just before resizing operations in letterbox.

One potential troubleshooting step is to ensure all images passed into the model are indeed of non-zero size after any preprocessing. It can sometimes help to explicitly check or print the shape of images at various points to confirm they're not being inadvertently altered to a zero-dimension array, which seems to be what OpenCV is complaining about (!dsize.empty()).

Considering the error happens during a cv2.resize call inside the letterbox function and your transformed array shape and dtype look correct, you might also want to inspect if there's an edge case where the preprocessing results in dimensions that OpenCV cannot handle (though this seems less likely given the shape you've provided).

Without altering much of your existing workflow and adding an explicit check, your code already appears well-structured for the task. So, I recommend inserting trouble-shooting print statements or debugging steps around the preprocessing steps, especially right before the model invocation, to ensure everything is as expected.

Let's keep debugging collaborative and insightful! If the issue persists or you find more specific patterns leading to the error, feel free to share more details. The YOLO community and the Ultralytics team are here to help. 🚀