I worked on an interesting project as a PHD student. It was rejected from a top conference with all but 1 reviewer agreeing that it was an ok paper but not good enough for that conference, because more work was required. The other reviewer said it was in top 15% papers. Do the reviewers mean "trash paper" when they say "ok paper" ? I don't have the time to finish it because I have switched my research area. In the project, I also collected and processed a large amount of data which would be useful for the computer vision community. Should I just put the paper and associated data/code on my website? Or should I put it on arxiv? In either cases, I wanted to acknowledge and include a link to the reviews from the conference. Is that a bad idea? The review process was double blind and hence the reviews are anonymous.
Definitely no. As Luke points out, since the reviews are blind, reviewers tend to be honest, if not leaning to the negative side. If they were explicit that the paper is fine, though still somewhat preliminary, understand it as a positive feedback.
I think the answer is yes, unless you want to "hide" the paper from the public.
I think about these things this way: the public (through taxes, NGO's, etc. and the subsequent redistribution to higher education and research) is paying us to do research with the hope that something good and beneficial will come out. Once you do the work, regardless of whether top-notch, or only somewhat significant, as far as it is a result of a serious effort and of reasonable quality (workmanship-wise), the public deserves reporting back on our work, as well as deserves the results. It is up to others to decide whether our work is useful to the society, or not.
Do both. Yes, going for arXiv is a good idea in such a case. If you feel so, you can prepend the paper with explanation of why you publish it as it is.
Either way, my advice is to publish the work. Actually, your situation is quite common. What many people do in a situation you describe, is to publish and present the work in a workshop associated with the (major) relevant conference. Or re-submit it as a poster, or short paper, if the conference has such a track. Those are definitely for quality, but still somewhat preliminary results.
No. I think you idea is good. Honest and open. How about including the reviews in an appendix of the arXiv submission?
Have you considered submission to a journal? As reviewers are not identified, it is reasonable to trust their judgement that the paper is "okay", and of course "okay" at a top conference in CS could still be quite good, especially as it's within the top 15%. Although perhaps the top-tier journals would render a similar judgement, there may be a venue where it is appropriate (I'm not familiar with the computer vision area), and you could publish without an excess of additional work.
Alongside this you can (usually) always put a preprint on ArXiv, few journals or conferences take issue with this now, which allows you to make it available in the meantime, regardless of what else you decide to do with it in the future.
However I would definitely not include the reviews. Although there is no rule against doing so, particularly as the reviewers shouldn't be identifiable, it contributes nothing to your research - the reader should be assessing the value themselves. The inclusion of reviews would only indicate that you want to boast about your work, or that you want people to accept it because someone else said it was good.
Don't feel bad if your paper was rejected, if you really want to publish, you could always submit it to another conference in its current form trying to address some of the concerns of the reviewers.
The problem with conferences (most) is that you don't get a second chance to submit it. There is nothing wrong with putting your paper out there in Arxiv, but remember that if it has not been published before, you could always suffer the danger of being plagiarized to certain extent.
Like Luke mentions, reviewers are most of the time a toss in the air.