Strip Pooling: Rethinking Spatial Pooling for Scene Parsing

By Qibin Hou, Li Zhang, Ming-Ming Cheng, Jiashi Feng. 2020

677 citations recorded by the source corpus.

Spatial pooling has been proven highly effective to capture long-range contextual information for pixel-wise prediction tasks, such as scene parsing. In this paper, beyond conventional spatial pooling that usually has a regular shape of NxN, we rethink the formulation of spatial pooling by introducing a new pooling strategy, called strip pooling, which considers a long but narrow kernel, i.e., 1xN or Nx1. Based on strip pooling, we further investigate spatial pooling architecture design by 1) introducing a new strip pooling module that enables backbone networks to efficiently model long-range dependencies; 2) presenting a novel building block with diverse spatial pooling as a core; and 3) systematically comparing the performance of the proposed strip pooling and conventional spatial pooling techniques. Both novel pooling-based designs are lightweight and can serve as an efficient plug-and-play modules in existing scene parsing networks. Extensive experiments on Cityscapes and ADE20K benchmarks demonstrate that our simple approach establishes new state-of-the-art results. Code is available at https://github.com/Andrew-Qibin/SPNet.

Subfields: Computer Vision and Pattern Recognition

How researchers see this paper

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

In Computer Vision and Pattern Recognition

Citations in the corpus

VOLO: Vision Outlooker for Visual Recognition (2021, neutral): “9 Strip Pooling [24] ResNet-101 ImgNet-1k 45.”
ALIFE: Adaptive Logit Regularizer and Feature Replay for Incremental Semantic Segmentation (2022, methodology): “Speciﬁcally, SDR minimizes distances between features of the same category [6], while PLOP distills intermediate features using a multi-scale strip pooling technique [8, 17].”
Representation Compensation Networks for Continual Semantic Segmentation (2022, methodology): “We conduct ablation MiB‡ [8] RC Strip [38] S-KD C-KD 15-1”
PLOP: Learning without Forgetting for Continual Semantic Segmentation (2020, methodology): “Note that while the first scale of Local POD (1/2) is equivalent to POD and models long-range dependencies, which are important for segmentation [64, 31, 53, 28], the subsequent scales (s = 1/2(1), 1/2(2) .”
Rethinking Semantic Segmentation from a Sequence-to-Sequence Perspective with Transformers (2020, neutral): “20 Strip pooling [23] 1K ResNet-101 54.”
Coordinate Attention for Efficient Mobile Network Design (2021, background): “Attention mechanisms [41, 40] have been proven helpful in a variety of computer vision tasks, such as image classification [18, 17, 44, 1] and image segmentation [14, 19, 10].”
Is Context-Aware CNN Ready for the Surroundings? Panoramic Semantic Segmentation in the Wild (2021, methodology): “Our concurrent attention module fundamentally differs from existing context aggregation methods [18], [29] as our attention aims to indicate which channels are critical at each individual column or row, instead of only aggregating the spatial dependency or reﬁning the features based on feature…”
Rotate to Attend: Convolutional Triplet Attention Module (2020, background): “Criss-Cross Networks (CCNet) [15] and SPNet [13] present novel attention blocks to capture rich contextual information using intersecting strips.”
Attention mechanisms in computer vision: A survey (2021, methodology): “[61] proposed strip pooling, a novel pooling method capable of encoding long-range context in either horizontal or vertical spatial domains.”
Scene Segmentation With Dual Relation-Aware Attention Network (2020, methodology): “In addition, we also perform better than the works modeling long-range dependencies with graph convolution [52] or strip pooling [54].”
SegNeXt: Rethinking Convolutional Attention Design for Semantic Segmentation (2022, background): “Thus, strip convolution can be a complement of grid convolutions and helps extract strip-like features [62, 29].”
Stripformer: Strip Transformer for Fast Image Deblurring (2022, background): “[22] exploit horizontal and vertical one-pixel long kernels to pool images and extract context information for scene parsing, called strip pooling, initially designed for contextual information extraction instead of deblurring.”
Improving Semantic Segmentation via Decoupled Body and Edge Supervision (2020, background): “To tackle the first problem, many approaches [6,7,8] have been proposed to enlarge the RF, such as dilated convolution, pyramid pooling module [8,9], non-local operators [10,11,12,13,14], graph convolution network (GCN) [15,16] and dynamic graph [17].”