Akshay Iyer
Akshay Iyer
Yeah, I had the same doubt regarding the image size of synthetic images not being 400*600. When I'm training on my own dataset, do I have to resize them to...
Interesting since it gives some good results. Is your paper published?
I deleted the files and downloaded them again into the same folder and ran the 2 python codes. But the error still persists. raw_groundtruth[0] contains multiple pose values but raw_groundtruth[1]...
Oh got the meaning of cleaning up. I removed the files and let the code generate it this time. That error went away but the length of train_dl is now...
Hey @mazatov , I'm curious to know what worked for you? I am facing a similar issue of SCAN loss plateauing and sometimes starting from a negative value and plateauing...
Also, @wvangansbeke, I loved this work and loved your implementation equally! I feel it's done beautifully, took me quiet sometime to understand the work. I have a couple of questions...
You could save the features in the mine_nearest_neighbors function. Those are your representations
It is already present in my file coco.py but still I'm getting the same error on running demo.ipynb. I tried changing the root directory path in both demo.ipynb as well...