Download presentation
Presentation is loading. Please wait.
Published byInge Yuwono Modified over 5 years ago
1
Accel: A Corrective Fusion Network for Efficient Semantic Segmentation on Video
CVPR 2019 Oral Samvit Jain; Xin Wang; Joseph E. Gonzalez University of California, Berkeley
2
Semantic Segmentation on Video Problem Definition
Input: a video clip, not any ground truth Output: segmentation of each frame
3
Semantic Segmentation on Video Classical Approach
Segment on each single frame
4
Accel, an efficient approach
“Cheap” feature extraction net → Resnet 18 “Expensive” feature extraction net → Resnet 101
5
Network Architecture Optical Flow Warp Operation
6
Experiment Ablation Study
𝑁 𝑅 is always ResNet 101
7
Experiment Accuracy vs. inference time
On CityScapes Dataset On CamVid Dataset
8
Experiment Comparison with Others
On CityScapes Dataset On CamVid Dataset
9
r1. input frames r2. Accel NR branch r3. Accel NU branch r4. NR+NU, Resnet18
10
Conclusions The structure is very simple.
And it looks faster and could get higher performance. But the comparison is unfair. The baseline is not STOA. Where is BiSeNet and ICNet??
Similar presentations
© 2025 SlidePlayer.com. Inc.
All rights reserved.