HerdNet results with a numpy array as input are not saveable through pw_utils.save_detection_images_dots

[calebrob6]

Search before asking

• I have searched the Pytorch-Wildlife issues and found no similar bug report.

Bug

Running the following:

img_path = "inputs/0_resized.png"

img = Image.open(img_path)
img = np.array(img)

results = detection_model.single_image_detection(img)
pw_utils.save_detection_images_dots(results, "outputs/", overwrite=False)

results in the following traceback:

---------------------------------------------------------------------------
FileNotFoundError                         Traceback (most recent call last)
Cell In[33], [line 7](vscode-notebook-cell:?execution_count=33&line=7)
      [4](vscode-notebook-cell:?execution_count=33&line=4) img = np.array(img)
      [6](vscode-notebook-cell:?execution_count=33&line=6) results = detection_model.single_image_detection(img)
----> [7](vscode-notebook-cell:?execution_count=33&line=7) pw_utils.save_detection_images_dots(results, "outputs/", overwrite=False)

File /opt/conda/envs/pytorch_wildlife/lib/python3.8/site-packages/PytorchWildlife/utils/post_process.py:123, in save_detection_images_dots(results, output_dir, input_dir, overwrite)
    [117](https://vscode-remote+ssh-002dremote-002bv1001.vscode-resource.vscode-cdn.net/opt/conda/envs/pytorch_wildlife/lib/python3.8/site-packages/PytorchWildlife/utils/post_process.py:117)         sink.save_image(
    [118](https://vscode-remote+ssh-002dremote-002bv1001.vscode-resource.vscode-cdn.net/opt/conda/envs/pytorch_wildlife/lib/python3.8/site-packages/PytorchWildlife/utils/post_process.py:118)             image=cv2.cvtColor(annotated_img, cv2.COLOR_RGB2BGR), image_name=image_name
    [119](https://vscode-remote+ssh-002dremote-002bv1001.vscode-resource.vscode-cdn.net/opt/conda/envs/pytorch_wildlife/lib/python3.8/site-packages/PytorchWildlife/utils/post_process.py:119)         )
    [120](https://vscode-remote+ssh-002dremote-002bv1001.vscode-resource.vscode-cdn.net/opt/conda/envs/pytorch_wildlife/lib/python3.8/site-packages/PytorchWildlife/utils/post_process.py:120) else:
    [121](https://vscode-remote+ssh-002dremote-002bv1001.vscode-resource.vscode-cdn.net/opt/conda/envs/pytorch_wildlife/lib/python3.8/site-packages/PytorchWildlife/utils/post_process.py:121)     annotated_img = lab_annotator.annotate(
    [122](https://vscode-remote+ssh-002dremote-002bv1001.vscode-resource.vscode-cdn.net/opt/conda/envs/pytorch_wildlife/lib/python3.8/site-packages/PytorchWildlife/utils/post_process.py:122)         scene=dot_annotator.annotate(
--> [123](https://vscode-remote+ssh-002dremote-002bv1001.vscode-resource.vscode-cdn.net/opt/conda/envs/pytorch_wildlife/lib/python3.8/site-packages/PytorchWildlife/utils/post_process.py:123)             scene=np.array(Image.open(results["img_id"]).convert("RGB")),
    [124](https://vscode-remote+ssh-002dremote-002bv1001.vscode-resource.vscode-cdn.net/opt/conda/envs/pytorch_wildlife/lib/python3.8/site-packages/PytorchWildlife/utils/post_process.py:124)             detections=results["detections"],
    [125](https://vscode-remote+ssh-002dremote-002bv1001.vscode-resource.vscode-cdn.net/opt/conda/envs/pytorch_wildlife/lib/python3.8/site-packages/PytorchWildlife/utils/post_process.py:125)         ),
    [126](https://vscode-remote+ssh-002dremote-002bv1001.vscode-resource.vscode-cdn.net/opt/conda/envs/pytorch_wildlife/lib/python3.8/site-packages/PytorchWildlife/utils/post_process.py:126)         detections=results["detections"],
    [127](https://vscode-remote+ssh-002dremote-002bv1001.vscode-resource.vscode-cdn.net/opt/conda/envs/pytorch_wildlife/lib/python3.8/site-packages/PytorchWildlife/utils/post_process.py:127)         labels=results["labels"],
    [128](https://vscode-remote+ssh-002dremote-002bv1001.vscode-resource.vscode-cdn.net/opt/conda/envs/pytorch_wildlife/lib/python3.8/site-packages/PytorchWildlife/utils/post_process.py:128)     )
    [129](https://vscode-remote+ssh-002dremote-002bv1001.vscode-resource.vscode-cdn.net/opt/conda/envs/pytorch_wildlife/lib/python3.8/site-packages/PytorchWildlife/utils/post_process.py:129)     sink.save_image(
    [130](https://vscode-remote+ssh-002dremote-002bv1001.vscode-resource.vscode-cdn.net/opt/conda/envs/pytorch_wildlife/lib/python3.8/site-packages/PytorchWildlife/utils/post_process.py:130)         image=cv2.cvtColor(annotated_img, cv2.COLOR_RGB2BGR), image_name=os.path.basename(results["img_id"])
    [131](https://vscode-remote+ssh-002dremote-002bv1001.vscode-resource.vscode-cdn.net/opt/conda/envs/pytorch_wildlife/lib/python3.8/site-packages/PytorchWildlife/utils/post_process.py:131)     )

File /opt/conda/envs/pytorch_wildlife/lib/python3.8/site-packages/PIL/Image.py:3431, in open(fp, mode, formats)
   [3428](https://vscode-remote+ssh-002dremote-002bv1001.vscode-resource.vscode-cdn.net/opt/conda/envs/pytorch_wildlife/lib/python3.8/site-packages/PIL/Image.py:3428)     filename = os.path.realpath(os.fspath(fp))
   [3430](https://vscode-remote+ssh-002dremote-002bv1001.vscode-resource.vscode-cdn.net/opt/conda/envs/pytorch_wildlife/lib/python3.8/site-packages/PIL/Image.py:3430) if filename:
-> [3431](https://vscode-remote+ssh-002dremote-002bv1001.vscode-resource.vscode-cdn.net/opt/conda/envs/pytorch_wildlife/lib/python3.8/site-packages/PIL/Image.py:3431)     fp = builtins.open(filename, "rb")
   [3432](https://vscode-remote+ssh-002dremote-002bv1001.vscode-resource.vscode-cdn.net/opt/conda/envs/pytorch_wildlife/lib/python3.8/site-packages/PIL/Image.py:3432)     exclusive_fp = True
   [3433](https://vscode-remote+ssh-002dremote-002bv1001.vscode-resource.vscode-cdn.net/opt/conda/envs/pytorch_wildlife/lib/python3.8/site-packages/PIL/Image.py:3433) else:

FileNotFoundError: [Errno 2] No such file or directory: '/home/davrob/ssdprivate/alaska-fish-and-game.caribou-counting/pw_test/None'

Environment

No response

Minimal Reproducible Example

No response

Additional

No response

Are you willing to submit a PR?

• Yes I'd like to help by submitting a PR!

[idchacon28]

Hello @calebrob6 ,

Thank you for bringing this to our attention. We have updated the HerdNet models to handle both image paths and numpy arrays. Please pull the latest changes from the repository to apply this update to your local copy. Your feedback is greatly appreciated as it helps us to improve the usability of our models. If you have any more issues or suggestions, please do not hesitate to share them.

I am going to close this issue now, but if you encounter any further problems or have additional feedback after updating, feel free to reopen it or create a new issue.

Thank you once again for your contribution!

Best regards,
@idchacon28