Yes, it sure does! My thumbs up, thumbs down detector does surprisingly well with just 30 images of each though. For actual use, much more data is needed.

This reminds me of Johnny Five. More input, More INPUT, MORE INPUT!

I have been working in Arduino today. I am tinkering with an LSM303 (Accelerometer and Magnetometer) and an NXP IMU (Accel, Mag, and Gyro). My goal is to get similar numbers for Roll, Pitch, and Heading from the NXP IMU as I get from the LSM303.

It is 3:14 am. I am feeling a bit irrational.

I am working on a lesson to classify just images of thumbs up or thumbs down. I need to get some sort of tripod for my Raspberry Pi camera so I can have it still while showing it my thumbs.

The model keeps wanting more data. I am now coming to appreciate what a good job Siri, Cortana, and Alexa actually do in recognizing speech from several different voices. I am supposed to collect at least 90 images each of thumbs up and thumbs down to get a reasonable prediction.

Um, well, er, more data pleaseā€¦


Yes, the second site should be Everything else is good.

I am just a poor Geek lost in a new decade. I should build a counter. I am knee-deep in AL (Artificial Learning). I do not believe a machine can ever be intelligent like us Humans. I am starting into image classification now. should point to my personal blog. Sorry for the confusion. should point to my technical blog. Sorry!

Could you please point at my personal blog? Please remove all references to


ACK! I lost 5.5GB of a 7.1GB download! I was running it in the background on an rPi3 so I could just let it run 24/7. All gone, starting over. This is all coming through my phone's hotspot.