Scaling Wide Residual Networks for Panoptic Segmentation

By Liang-Chieh Chen, Huiyu Wang, Siyuan Qiao. 2020

54 citations recorded by the source corpus.

The Wide Residual Networks (Wide-ResNets), a shallow but wide model variant of the Residual Networks (ResNets) by stacking a small number of residual blocks with large channel sizes, have demonstrated outstanding performance on multiple dense prediction tasks. However, since proposed, the Wide-ResNet architecture has barely evolved over the years. In this work, we revisit its architecture design for the recent challenging panoptic segmentation task, which aims to unify semantic segmentation and instance segmentation. A baseline model is obtained by incorporating the simple and effective Squeeze-and-Excitation and Switchable Atrous Convolution to the Wide-ResNets. Its network capacity is further scaled up or down by adjusting the width (i.e., channel size) and depth (i.e., number of layers), resulting in a family of SWideRNets (short for Scaling Wide Residual Networks). We demonstrate that such a simple scaling scheme, coupled with grid search, identifies several SWideRNets that significantly advance state-of-the-art performance on panoptic segmentation datasets in both the fast model regime and strong model regime.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition, Model Architectures

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Computer Vision and Pattern Recognition

In Model Architectures

Citations in the corpus

OneFormer: One Transformer to Rule Universal Image Segmentation (2022, neutral): “1 Panoptic-DeepLab‡ [11] SWideRNet† [8] 536M 10365G — 1025×2049 60k 66.”
Masked-attention Mask Transformer for Universal Image Segmentation (2021, methodology): “…mIoU (s.s.) (m.s.) Panoptic FCN [31] Swin-L † 65.9 - - - - Panoptic-DeepLab [11] SWideRNet [9] 66.4 40.1 82.2 - - Panoptic-DeepLab [11] SWideRNet [9] 67.5 ∗ 43.9 ∗ 82.9 ∗ - - SETR [64] ViT-L † nal predictions of Mask2Former after the Transformer de-coder layers, i.e ., layer 9, and AR@100…”
CMT-DeepLab: Clustering Mask Transformers for Panoptic Segmentation (2022, methodology): “Additionally, we obtain a larger network backbone by scaling up the number of blocks in stage-4 of the backbone [14].”
k-means Mask Transformer (2022, methodology): “…k MaX-DeepLab, with single-scale testing, shows a signiﬁcant improvement of 1.4% PQ compared to the previous state-of-art Panoptic-DeepLab [22] with SWideRNet-(1, 1, 4.5) [16] as backbone, which adopts multi-scale testing, resulting in over 60 × more computational costs compared to k MaX-DeepLab.”