1 code implementation • 14 Sep 2022 • Kirill Prokofiev, Vladislav Sovrasov
Although transformer-based heads are considered to achieve better results than graph-based branches, we argue that with the proper training strategy, graph-based methods can demonstrate just a small accuracy drop, while spending less computational resources on inference.
Ranked #2 on Multi-Label Classification on NUS-WIDE
1 code implementation • 16 Aug 2021 • Kirill Prokofiev, Vladislav Sovrasov
Nowadays deep learning-based methods have achieved a remarkable progress at the image classification task among a wide range of commonly used datasets (ImageNet, CIFAR, SVHN, Caltech 101, SUN397, etc.).
2 code implementations • 23 Jun 2021 • Ilya Krylov, Sergei Nosov, Vladislav Sovrasov
A large scale human-labeled dataset plays an important role in creating high quality deep learning models.
1 code implementation • Submitted to International Conference on Pattern Recognition (ICPR 2020) 2020 • Vladislav Sovrasov, Dmitry Sidnev
This work considers the problem of domain shift in person re-identification. Being trained on one dataset, a re-identification model usually performs much worse on unseen data.
Ranked #31 on Person Re-Identification on MSMT17