neilyoung

Hi @Namburger, I was now trying to retrain the whole model, to no avail. It stops with `./retrain_detection_model.sh: line 50: 59 Killed` after a short time. ~~I'm now considering to...

OK, I was trying to follow your Medium article, but I got stuck on the demo: 

Really strange, I omitted the "width" parameter from the draw.rectangle call and it works. Strange, because I think, the parameter is correct... BTW: Python 3.6

I was running your tutorial. It took 7 hours on the colab server and failed finally in step 17 and following. `ERROR: tflite_runtime-2.5.0-cp36-cp36m-linux_x86_64.whl is not a supported wheel on this...

I regret that I have to talk to myself only, but message to myself: I guess this would be the correct call `!edgetpu_compiler -s ssdlite_mobiledet_dog_vs_cat.tflite ` instead of `!edgetpu_compiler -s...

Well, after the change I got an "edgetpu.tflite" file and I could confirm your results. 8 images, 0.15 s (if I interpret the output correclty). I somehow had a couple...

Created a PR regarding the wrong name https://github.com/Namburger/edgetpu-ssdlite-mobiledet-retrain/pull/7 I hope I'm not wrong with this.

Besides this Colab excursion - are there other hints regarding how to make the initial tutorial run? I can't achieve any detection, not with full training (stops), not with changed...

Well, that was easy. But anyway - the spp crashes

Comment the include statement. Not used. In any case - the app doesn't work any longer. Crashes in my case