DFANet: Deep Feature Aggregation for Real-Time Semantic Segmentation

By Hanchao Li, Pengfei Xiong, Haoqiang Fan. Computer Vision and Pattern Recognition, 2019

676 citations recorded by the source corpus.

This paper introduces an extremely efficient CNN architecture named DFANet for semantic segmentation under resource constraints. Our proposed network starts from a single lightweight backbone and aggregates discriminative features through sub-network and sub-stage cascade respectively. Based on the multi-scale feature propagation, DFANet substantially reduces the number of parameters, but still obtains sufficient receptive field and enhances the model learning ability, which strikes a balance between the speed and segmentation performance. Experiments on Cityscapes and CamVid datasets demonstrate the superior performance of DFANet with 8$\times$ less FLOPs and 2$\times$ faster than the existing state-of-the-art real-time semantic segmentation methods while providing comparable accuracy. Specifically, it achieves 70.3\% Mean IOU on the Cityscapes test dataset with only 1.7 GFLOPs and a speed of 160 FPS on one NVIDIA Titan X card, and 71.3\% Mean IOU with 3.4 GFLOPs while inferring on a higher resolution image.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition

View Paper: https://www.semanticscholar.org/paper/c2aa0581254ce342c77b2c7a835b3587bd453486

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

Citations in the corpus

Deep Dual-Resolution Networks for Real-Time and Accurate Semantic Segmentation of Traffic Scenes (2023, methodology): “Images are randomly cropped into 1024 × 1024 for training following [17], [22], [36].”
BiSeNet V2: Bilateral Network with Guided Aggregation for Real-Time Semantic Segmentation (2020, background): “…al., 2016), ESPNet (Mehta et al., 2018), ESPNetV2 (Mehta et al., 2019), ERFNet (Romera et al., 2018), Fast-SCNN (Poudel et al., 2019), ICNet (Zhao et al., 2018a), DABNet (Li et al., 2019a), DFANet (Li et al., 2019b), GUN (Mazzini, 2018), SwiftNet (Or-sic et al., 2019), BiSeNetV1 (Yu et al., 2018a).”
PIDNet: A Real-time Semantic Segmentation Network Inspired by PID Controllers (2022, methodology): “DFANet [27] introduced a light-weight backbone by modifying the architecture of Xception [11], which was based on depth-wise separable convolution, and reduced the input size for faster inference speed.”
Class-Incremental Learning for Semantic Segmentation Re-Using Neither Old Data Nor Old Labels (2020, background): “Recently, a large interest in developing extremely efficient architectures arised [16], [17], [18], [19], [20].”
HyperSeg: Patch-wise Hypernetwork for Real-time Semantic Segmentation (2020, methodology): “Subsequent methods beneﬁted from progress in efﬁcient network architecture [24, 51], such as depth-wise separable convolutions [7, 20] and inverted residual blocks [40] which we also use in our work.”
TopFormer: Token Pyramid Transformer for Mobile Semantic Segmentation (2022, methodology): “DFANet [24] utilizes a light-weight backbone to speed up its network and proposes a cross-level feature aggregation to boost accuracy.”
Lite-HRNet: A Lightweight High-Resolution Network (2021, methodology): “02 GFLOPs, outperforming the hand-crafted methods [34, 55, 50, 23, 58, 36, 41, 12] and NAS-based methods [54, 25, 16, 24] , and comparable with SwiftNetRN18 [32] that is far computationally intensive (104 GFLOPs).”
EfficientViT: Lightweight Multi-Scale Attention for High-Resolution Dense Prediction (2023, background): “In addition, there are also some works targeting improving the efficiency of high-resolution dense prediction models [57, 40, 29, 52].”
Rethinking BiSeNet For Real-time Semantic Segmentation (2021, methodology): “DFANet [18] adopted a lightweight backbone to reduce computation cost and devised a cross-level feature aggregation module to enhance performance.”