Hello! I recently got an OAK-D camera for our project we need to find the distance between our camera and the object we're trying to detect. Our app is currently running the model we trained locally but the FPS are too low even with GPU so I want to try running directly on the oak camera. Following are some details about my model.
It's a YOLOv8 model but I also tried with YOLOv5. It only has one class and the images used are also very few (around 50). I used the online tools to get the blob file like. Image shapes tried both 640 and 416.
https://tools.luxonis.com/
https://blobconverter.luxonis.com/
I'm having trouble running these blob files, here's what I tried.
Putting the blob model data in the models folder of depthai repo (recourses/nn/custommodel) and ran the depthai_demo.py. With the default models it works fine but when I select my custom model, there it gives me following error
Original message 'Couldn't read data from stream: 'rgb' (X_LINK_ERROR)'
I also tried doing it manually from what I could find but there it either gives me same error as above or another error
Mask is not defined for output {LAYER NUMBER}
I can't figure out what's the issue, whether I need to fix my model or its something else. training the model from start isn't an issue either but the official guide uses older versions of TensorFlow so I've had a lot of trouble running those notebooks. Anything to help would be appreciated!
