Title | ||
---|---|---|
CReaM: Condensed Real-time Models for Depth Prediction using Convolutional Neural Networks |
Abstract | ||
---|---|---|
Since the resurgence of CNNs the robotic vision community has developed a range of algorithms that perform classification, semantic segmentation and structure prediction (depths, normals, surface curvature) using neural networks. While some of these models achieve state-of-the art results and super human level performance, deploying these models in a time critical robotic environment remains an ongoing challenge. Real-time frameworks are of paramount importance to build a robotic society where humans and robots integrate seamlessly. To this end, we present a novel real-time structure prediction framework that predicts depth at 30 frames per second on an NVIDIA-TX2. At the time of writing, this is the first piece of work to showcase such a capability on a mobile platform. We also demonstrate with extensive experiments that neural networks with very large model capacities can be leveraged in order to train accurate condensed model architectures in a “from teacher to student” style knowledge transfer. |
Year | DOI | Venue |
---|---|---|
2018 | 10.1109/IROS.2018.8594243 | 2018 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) |
Keywords | DocType | Volume |
CNNs,robotic vision community,semantic segmentation,surface curvature,robotic society,real-time structure prediction framework,NVIDIA-TX2,CReaM,real-time models,depth prediction,convolutional neural networks,classification,mobile platform,condensed model architectures | Conference | abs/1807.08931 |
ISSN | ISBN | Citations |
2153-0858 | 978-1-5386-8095-7 | 3 |
PageRank | References | Authors |
0.37 | 5 | 3 |
Name | Order | Citations | PageRank |
---|---|---|---|
Andrew Spek | 1 | 4 | 1.40 |
Thanuja Dharmasiri | 2 | 5 | 0.72 |
Tom Drummond | 3 | 2676 | 159.45 |