(AF)2-S3Net: Attentive Feature Fusion with Adaptive Feature Selection for Sparse Semantic Segmentation Network

By Ran Cheng, Ryan Razani, Ehsan Moeen Taghavi, Enxu Li, Bingbing Liu. Computer Vision and Pattern Recognition, 2021

310 citations recorded by the source corpus.

Autonomous robotic systems and self driving cars rely on accurate perception of their surroundings as the safety of the passengers and pedestrians is the top priority. Semantic segmentation is one of the essential components of road scene perception that provides semantic information of the surrounding environment. Recently, several methods have been introduced for 3D LiDAR semantic segmentation. While they can lead to improved performance, they are either afflicted by high computational complexity, therefore are inefficient, or they lack fine details of smaller object instances. To alleviate these problems, we propose (AF)2-S3Net, an end-to-end encoder-decoder CNN network for 3D LiDAR semantic segmentation. We present a novel multibranch attentive feature fusion module in the encoder and a unique adaptive feature selection module with feature map re-weighting in the decoder. Our (AF)2-S3Net fuses the voxel-based learning and point-based learning methods into a unified framework to effectively process the potentially large 3D scene. Our experimental results show that the proposed method outperforms the state-of-the-art approaches on the large-scale nuScenes-lidarseg and SemanticKITTI benchmark, ranking 1st on both competitive public leaderboard competitions upon publication.

Subfields: Aerospace Engineering, Computer Vision and Pattern Recognition

View Paper: https://www.semanticscholar.org/paper/dd2c89052f4dfb1f1d060b08374469aa1e0f57a6

How researchers see this paper

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

In Aerospace Engineering

1. Zhengbiao Zhu
2. Chaoda Zheng
3. Jiwen Lu
4. Xu Yan
5. Yunpeng Zhang
6. Jie Zhou
7. Shiv Ram Dubey
8. Jingkuan Song

In Computer Vision and Pattern Recognition

1. Jie Zhou
2. Jiwen Lu
3. Zhengbiao Zhu
4. Chaoda Zheng
5. Shenghui Cui
6. Xu Yan
7. Yi Wei
8. Yunpeng Zhang

Citations in the corpus

• SurroundOcc: Multi-Camera 3D Occupancy Prediction for Autonomous Driving (2023, methodology): “It has empowered the success of numerous methods on the lidar segmentation [32, 51, 12, 61, 60] and 3D scene completion [8, 45, 10, 26, 58] tasks.”
• Transformer-Based Visual Segmentation: A Survey (2023, unknown): “Foroutdoorscenes,point cloudsegmentationcanbedividedintopoint-based[100],[106] andvoxel-based[107],[108]approaches.”
• RPVNet: A Deep and Efficient Range-Point-Voxel Fusion Network for LiDAR Point Cloud Segmentation (2021, background): “AF2S3Net [8] stands on the shoulder of [10] and achieves the state-of-the-art of past methods, which proposes two novel attention blocks named Attentive Feature Fusion Module(AF2M) and Adaptive Feature Selection Module(AFSM), to effectively learn local and global contexts and emphasize the fine deta”
• 2DPASS: 2D Priors Assisted Semantic Segmentation on LiDAR Point Clouds (2022, methodology): “AF(2)-S3Net [41] applies multiple branches with different kernel sizes, aggregating multi-scale features via an attention mechanism.”
• Tri-Perspective View for Vision-Based 3D Semantic Occupancy Prediction (2023, background): “The ability to describe fine-grained 3D structures makes voxel-based representation favorable for 3D semantic occupancy prediction tasks including lidar segmentation [12, 29, 40, 44, 45, 51] and 3D scene completion [5, 10, 23, 38, 43].”