No description, website, or topics provided.
Switch branches/tags
Nothing to show
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Type Name Latest commit message Commit time
Failed to load latest commit information.
images
LICENSE
README.md
camera-recog.py
sample.txt
synset.txt

README.md

Image/camera input recognition using pre-trained models

A python demo code for image/camera input recognition using pre-trained models in Chainer by Shizuo KAJI

Licence

MIT Licence

Requirements

  • python 3.6

Install Anaconda if you do not have python 3 on your system.

  • Chainer > 2: pip install chainer

If you'd like to use GPU, cupy with CUDA support is additionally required.

  • OpenCV 3: Use Homebrew on mac, or download the binary from here and pip on Windows to install.

Example

python camera-recog.py -h for a brief description of command line arguments

python camera-recog.py opens the camera of your PC and outputs recognised objects captured by it. Hit 'p' to pause/resume, 'q' to quit.

python camera-recog.py -i sample.txt -n 6 scans the text file sample.txt and outputs 6 guesses for each image specified in the file. Sample images are obtained from Wikipedia.

python camera-recog.py -i sample.txt -n 5 -a vgg -g 0 same as above but with the VGG model and GPU.