andreapiso

Did you figure that out? would be interesting to me as well

I am retraining yolov2 on VOC 2012 with 20 classes and did not change any parameter. Loss is now at 0.01 and still cannot see any bounding box after 7000...

Please note that this app does not detect faces. It "just" does landmarking. The face detection is done by the standard Apple library. To answer your question, of course you...

Also, you should zip the file to see how much space it will actually take when you build the app, since the IPA is already a compressed archive.

Is this an issue of installing from different conda channels? My `pyarrow` comes from conda default and turbodbc from `conda-forge`. It seems `turbodbc` does not see `pyarrow` even if `pyarrow`...

I inserted this functionality and fixed a few bugs related to this and set up a model to train on VOC. Do you want a PR?

@RyanMorris95 do you still need that? I cannot just post the code as I made a lot of changes but I can help... what have you tried so far?

Standard training already uses pre-trained SqueezeNet trained on Imagenet to start, you should be able to use your 1060 if you are patient enough... especially since the GPU is starved...

Download the one from classification, not the one from detection, the detection model is based on the classification squeeze net so it is a good way to initialise it. You...

both in training and demo the resize function distorts the image, which cannot be good for learning... probably would be better to rescale keeping the aspect ratio and add padding?...