Re: Question regarding the experiments in this year's CVPR paper
Vladimir Bychkovsky
I've added files with image IDs for training and test sets to the
paper webpage[1]. These numbers were randomly generated, in other
words, the results should not vary if you use a different set.
Vlad.
[1] http://people.csail.mit.edu/vladb/photoadjust/
On Tue, Aug 16, 2011 at 5:39 AM, Sung Ju Hwang <t-su...@microsoft.com> wrote:
> Hi Vladimir,
>
>
>
> I’m Sung Ju Hwang from UT Austin. This summer I’m working as an intern at
> MSR, on automatic image enhancement project, with Dr. Sing Bing Kang and
> Ashish Kapoor.
>
> Now we’re working with your dataset, and want to compare the results with
> your method, but we do not know which training/test image to use, and random
> selection doesn’t seem like a way to do a fair comparison.
>
> I’m wondering if you could provide us with the index of the training and
> test images, so we could test our method on those test set?
>
> I thank you so much in advance and hope you have a great day.
>
>
>
> Best Regards,
>
> Sung Ju
>
>
>
>