BLOB SIZE IS TOO BIG ERROR

dhunjoshi

Hello
I trained custom object detection model as per colab notebook provided , but when i run the model in megaAI error come blob size is too big and its continuously run in loop everytiem shows same message.

Kindly help.

Brandon

Hi dhunjoshi ,

Sorry about the trouble here. Which network did you train?

So if it's MobileNetSSD or YOLO, could you try running in the following notebooks instead?

https://docs.luxonis.com/projects/api/en/gen2_develop/samples/08_rgb_mobilenet/
https://docs.luxonis.com/projects/api/en/gen2_develop/samples/22_tiny_tolo_v3_decoding_on_device/
If it's neither of those could you let us know which and then we can give some other examples.

The reason to do Gen2 is we made larger models supported by default in this Gen2 API. There is a more dynamic resource allocator.

Also can if you can share your model (feel free to email it to support@luxonis.com) we can try running it as well.

Thoughts?

Thanks,
Brandon

dhunjoshi

Brandon Thank for your reply . it is mobileNetSSD. I will share the model .

Regards

Brandon

Great thanks!

dhunjoshi

Brandon Hell brandon.
Thank you very much for your support in every post question. As i asked in earlier post i am working in a ocr project. i tried in mobileNetSSD but i did not get good result. i try with yolov5 and i got good result. Can you help me is it possible to convert yolov5 to megaAI format with same accuracy. i converted yolov5 model to onnx and it is working fine .

thanks and regards
Ganesh

Brandon

Hi @dhunjoshi ,

Yes, definitely. We have not yet gotten it to work yet either. We have gotten YOLOv3 and YOLOv4 to work, but haven't yet dug into V5. But we will give it a shot.

Thanks,
Brandon