Implementing fast semantic segmentation with CPU alone (DeeplabV3)
[4-5 FPS / Core m3 CPU only] [11 FPS / Core i7 CPU only] OpenVINO+DeeplabV3 RealTime semantic-segmentaion. Python3.5+Tensorflow v1.11.0+OpenCV3.4.3+PIL Latte Panda Alpha or Other x64 PC ...learn more
Project status: Under Development
Internet of Things, Artificial Intelligence
Groups
Internet of Things,
DeepLearning
Intel Technologies
OpenVINO,
Intel Opt ML/DL Framework,
Movidius NCS
Overview / Usage
- Environment
- LattePanda Alpha (Intel 7th Core m3-7y30) or LaptopPC (Intel 8th Core i7-8750H)
- Ubuntu 16.04 x86_64
- OpenVINO toolkit 2018 R4 (2018.4.420) or Any further
- Python 3.5
- OpenCV 3.4.3
- PIL
- Tensorflow v1.11.0 or Tensorflow-GPU v1.11.0 (pip install)
- DeeplabV3 + MobilenetV2 (Pascal VOC 2012)
- USB Camera (PlaystationEye) / Movie file (mp4)
- Benchmark
https://ncsforum.movidius.com/discussion/1329/lattepanda-alpha-openvino-cpu-core-m3-vs-ncs1-vs-ncs2-performance-comparison
- Usage
https://github.com/PINTO0309/OpenVINO-DeeplabV3#usage
- Future tasks
- Work that makes compatibility with NCS and NCS2.
- Work that makes compatibility with Intel GPU.
Methodology / Approach
OpenVINO offloads unsupported layers to Tensorflow and performs segmentation fast.
Technologies Used
OpenVINO+DeeplabV3
FP32
Not currently compatible with NCS and NCS2.
Repository
https://github.com/PINTO0309/OpenVINO-DeeplabV3
Other links
Collaborators
There are no people to show.