PYNQ-Z2 Dev Kit - CIFAR-10 Webcam continued....

RE: PYNQ-Z2 Dev Kit - CIFAR-10 Webcam continued....

dubbie — Thu, 15 Aug 2019 08:31:37 GMT

It is one of the strange (?) things about using Artificial Neural Networks (ANN), particularly when using image systems or other data that is viewed by humans, is that they use all the data. When a human looks at the car picture they just see the car and use that part of the image for recognition. an ANN isn't intelligent and uses all the image, including the background so when recognising the car, or plane or whatever, it is actually just recognising the array of pixel values and working out the nearest trained image of pixels values it has been given. So by making all the backgrounds similar (blank or anything else really) then it becomes the differences such as the car or plane and so on, that the ANN uses to make a match. The ANN will also use the position of the different pixels as well, so either the car or plane has to be precisely centred in every image, or, the training data must contain images of the car located in as many alternative locations within the image as possible. It can lead to a very large training data set, with the corresponding problem that the larger the training data set then the longer the training period, along with a greater possibility of not converging onto a working solution. Another problem with ANNs is, depending on the type of ANN and training used, that because training starts with some randomised numbers that training is not repeatable. So you could train, get a working solution, decide to retrain without changing anything and then when retraining, it will just not converge. My approach was that if a working system was obtained, always make sure to keep a copy of it before retraining.

Dubbie

RE: PYNQ-Z2 Dev Kit - CIFAR-10 Webcam continued....

clem57 — Thu, 15 Aug 2019 04:46:58 GMT

Training is more art than science[View:/resized-image/__size/16x16/__key/commentfiles/f7d226abd59f475c9d224a79e3f0ec07-9bda5d5e-81ad-4c03-8e7c-656d079d0761/contentimage_5F00_4722.png:16:16]

RE: PYNQ-Z2 Dev Kit - CIFAR-10 Webcam continued....

genebren — Wed, 14 Aug 2019 14:50:24 GMT

Ralph,

Interesting results. The models really seem to be heavily weighted to call everything airplanes (even a plastic elk). Image quality, likeness of image to learning images and many other minor details seem to influence the item classification.

Great work on digging in and comparing these classifiers.

Gene

RE: PYNQ-Z2 Dev Kit - CIFAR-10 Webcam continued....

dubbie — Wed, 14 Aug 2019 12:27:25 GMT

I agree with the comment about not being too good in the real world. Artificial Neural Networks can work well when the data set is limited or clearly distinguishable but once the data gets a bit iffy and vague they are not so good. They can be fun things to play with thou.

Dubbie