Multi-Exit Semantic Segmentation Networks

06/07/2021
by   Alexandros Kouris, et al.
0

Semantic segmentation arises as the backbone of many vision systems, spanning from self-driving cars and robot navigation to augmented reality and teleconferencing. Frequently operating under stringent latency constraints within a limited resource envelope, optimising for efficient execution becomes important. To this end, we propose a framework for converting state-of-the-art segmentation models to MESS networks; specially trained CNNs that employ parametrised early exits along their depth to save computation during inference on easier samples. Designing and training such networks naively can hurt performance. Thus, we propose a two-staged training process that pushes semantically important features early in the network. We co-optimise the number, placement and architecture of the attached segmentation heads, along with the exit policy, to adapt to the device capabilities and application-specific requirements. Optimising for speed, MESS networks can achieve latency gains of up to 2.83x over state-of-the-art methods with no accuracy degradation. Accordingly, optimising for accuracy, we achieve an improvement of up to 5.33 pp, under the same computational budget.

READ FULL TEXT

page 15

page 16

page 17

research
10/24/2022

SphNet: A Spherical Network for Semantic Pointcloud Segmentation

Semantic segmentation for robotic systems can enable a wide range of app...
research
01/30/2023

SeaFormer: Squeeze-enhanced Axial Transformer for Mobile Semantic Segmentation

Since the introduction of Vision Transformers, the landscape of many com...
research
02/02/2021

It's always personal: Using Early Exits for Efficient On-Device CNN Personalisation

On-device machine learning is becoming a reality thanks to the availabil...
research
07/13/2022

SlimSeg: Slimmable Semantic Segmentation with Boundary Supervision

Accurate semantic segmentation models typically require significant comp...
research
10/01/2019

Real-Time Semantic Stereo Matching

Scene understanding is paramount in robotics, self-navigation, augmented...
research
03/09/2020

FarSee-Net: Real-Time Semantic Segmentation by Efficient Multi-scale Context Aggregation and Feature Space Super-resolution

Real-time semantic segmentation is desirable in many robotic application...
research
05/02/2019

Approximate LSTMs for Time-Constrained Inference: Enabling Fast Reaction in Self-Driving Cars

The need to recognise long-term dependencies in sequential data such as ...

Please sign up or login with your details

Forgot password? Click here to reset