Tomasz Malisiewicz

Thanks a lot for looking into this!

The API will also include a monolithic mode of training, as in DalalTriggs-style detector.

almost done...

Almost done!

I don't think those annotations are still around. There were something like 600 buses in total and I made a simple Matlab GUI for clicking on points. It should take...

What happens is one of the nodes is waiting for files to finish, while another process deletes them, so the first process is stuck in an infinite loop waiting for...

I wrote the `esvm_features` function so that given a [MxNx3] image, it returns the HOG feature vector of size [M/sbin,N/sbin,F]. ``` matlab features = esvm_features(I,sbin) ``` To encode the dimensionality...

You'll have to look closely at the code: by default I was employing a method which is more powerful than Platt's calibration method. This method involves a "boosting" matrix which...

HOG is not typically used for image classification tasks, but instead object detection tasks. When applying dense SIFT or GIST (or engineering your own feature) for image classification, you have...

Hi Angela, That's cool that you figured out how to change the esvm code to work for your object recognition scenario. Even though sliding window is often not needed for...