Review on Neural Compute Stick 2 by Intel by Nick Griffith

I've had it for a little over a week and have been able to convert a custom TensorFlow model into their framework, but can ONLY use it in my demo programs. All in all, I'm more than disappointed because I'm Intel. Some of the requirements they make are just not acceptable if you plan to use your machine for anything other than these products. Special versions of OpenCV (without using environments!) I was able to get pre-configured "examples" of these, which are binaries, so there's no point in using this product for your own needs. Support and forums are what you would expect from a large company like Intel: 0 Help for all "Can this thread be closed". They have multiple editions, but none of them that I've found are plug and play like they suggest. Python libraries are not loaded. I'd love to talk about the documentation, but other than being able to install and run the demo, there's nothing I can do that renders it useless. Thought I'd buy one of these as all Coral USB sticks are on hold. If you're in the same boat as me, do yourself a favor and just wait for your Coral Sticks. You can always add more data to train your model! UPDATE: After going through all the github and support requests I was able to get working output with my own custom mockup. Searching for multiple answers I found even more issues with their product and lack of guidance on some important aspects of their software. From array input ordering to blob restructuring, this seems like one of those products that you can only use if you never plan on updating anything in the box. I was able to see impressive improvements in my FPS of 0.32 with the TFLite model compared to 1.89 with this joystick. I do not advise against using it YET due to the lack of support.