cv2.imdecode(image, cv2.IMREAD_COLOR) doesn't work

[imanshahmari]

in simple-infer.py line 55. it returns null which can not later be shown in cv2.imshow('image', image)
Error:

cv2.error: OpenCV(4.6.0) /Users/xperience/actions-runner/_work/opencv-python/opencv-python/opencv/modules/highgui/src/window.cpp:967: error: (-215:Assertion failed) size.width>0 && size.height>0 in function 'imshow

[mtraor3]

Hi @imanshahmari - I just tried the code with my object detection model and didn't have issues. What did you set for ROBOFLOW_SIZE in your roboflow_config.json file?

Additionally, did you select the correct input for your camera for cv2.VideoCapture(0) before the function definition? Sometimes, external cameras plugged in by USB will be set as device 1 or 2 rather than 0 (depending on the number of cameras you have connected to your system).

• e.g, cv2.VideoCapture(1) or cv2.VideoCapture(2)

[idwessough]

Same issue here the API respond "message: Not found" when passing an image which is encoded as snippets do.

[deBrian07]

Same issue here, it seems like this issue is caused by passing an empty array to imshow function, but not sure why it's empty.
Error:

cv.imshow('image', image)
cv2.error: OpenCV(4.6.0) /tmp/opencv-20221208-24159-49apm1/opencv-4.6.0/modules/highgui/src/window.cpp:967: error: (-215:Assertion failed) size.width>0 && size.height>0 in function 'imshow'

Please fix it ASAP, thanks!

[mtraor3]

@imanshahmari @idwessough @deBrian07 for this example dataset: https://universe.roboflow.com/mohamed-traore-2ekkp/face-detection-mik1i/model/15, I've successfully deployed it to my webcam.

In your roboflow_config.json file, you'll want to edit "ROBOFLOW_MODEL" to the model_id/version#

• example with the dataset linked above: "ROBOFLOW_MODEL": "face-detection-mik1i/15"

Additionally, include your Private API Key for "ROBOFLOW_API_KEY"

• example (your own dataset, in your own workspace): Accessing Your API Key
• example (Roboflow Universe pre-trained model): You will find the private API key for deployment under the Model tab, when you scroll down. I've cut off the screenshot to the point right above where my API key is, as you don't want to reveal your Private API key to others -

For the "ROBOFLOW_SIZE" parameter in the roboflow_config.json file, set the value to what you selected for Resize in Preprocessing

• In the dataset I've linked above, I set Resize to 640 x 640 when I generated version 15, so my value would look like this: "ROBOFLOW_SIZE": 640

From here, things should run fine. If you receive an error about the device not being available, then be sure you have allowed for your webcam to be accessed by the program you are running the script with. If it still errors, it is likely the device number needs to be updated. Sometimes, external cameras plugged in by USB will be set as device 1 or 2 rather than 0 (depending on the number of cameras you have connected to your system).

• e.g, video = cv2.VideoCapture(1) or video = cv2.VideoCapture(2) in line 31 of infer_simple.py, rather than video = cv2.VideoCapture(0)

I've included the sample config file for my Face Detection project I linked above. Again, just be sure to update the ROBOFLOW_API_KEY, the ROBOFLOW_MODEL, and ROBOFLOW_SIZE parameters to match the values for your dataset.

• Sample: roboflow_config.json

{
    "__comment1": "Obtain these values via Roboflow",
    "ROBOFLOW_API_KEY": "INSERT_PRIVATE_API_KEY",
    "ROBOFLOW_MODEL": "face-detection-mik1i/15",
    "ROBOFLOW_SIZE": 640,

    "__comment2": "The following are only needed for infer-async.py",
    "FRAMERATE": 24,
    "BUFFER": 0.5
}