Awesome

RDI-Net

This contains the PyTorch implementation of the RDI-Net papers,

RDI-Net: Relational Dynamic Inference Networks, ICCV 2021.

By Huanyu Wang, Songyuan Li, Shihao Su, Zequn Qin, Xi Li.

Abstract

Dynamic inference networks, aimed at promoting computational efficiency, go along an adaptive executing path for a given sample. Prevalent methods typically assign a router for each convolutional block and sequentially make block-by-block executing decisions, without considering the relations during the dynamic inference. In this paper, we model the relations for dynamic inference from two aspects: the routers and the samples. We design a novel type of router called the relational router to model the relations among routers for a given sample. In principle, the cur- rent relational router aggregates the contextual features of preceding routers by graph convolution and propagates its router features to subsequent ones, making the executing decision for the current block in a long-range manner. Fur- thermore, we model the relation between samples by intro- ducing a Sample Relation Module (SRM), encouraging cor- related samples to go along correlated executing paths. As a whole, we call our method the Relational Dynamic Inference Network (RDI-Net). Extensive experiments on CIFAR- 10/100 and ImageNet show that RDI-Net achieves state-of- the-art performance and computational cost reduction.

Paper

Usage

python main_dist.py --dataset cifar10 --train_bs 256 --test_bs 100 --epochs 320 --lr 0.1 --model R110_C10 --weight 0.0 --log_path ./outputs/ --note REBUTTAL

License and Citation

@inproceedings{wang2021RDINET,
    author = {Huanyu Wang and Songyuan Li and Shihao Su and Zequn Qin and Xi Li},
    title = {RDI-Net: Relational Dynamic Inference Networks},
    booktitle="Proc. IEEE Int. Conf. Comput. Vis.",
    year = {2021},
}

Disclamer

We based our code on Convnet-AIG, CoDiNet, please go show some support!