Skip to content

Latest commit

 

History

History
68 lines (57 loc) · 1.89 KB

README.md

File metadata and controls

68 lines (57 loc) · 1.89 KB

mlAlgorithms

This repo contains some basic ML models like Alexnet,VGG,Resnet...
Package tensorflow contains these models implementations using tensorflow >= 1.0.
Package mxnet contains these models implementations using mxnet >= 0.10.

Indeed,all the models can be found at model zoom of relative frameworks like tensorflow,mxnet

Warn

I have worked with mxnet,so I can maintain the package mxnet continually.Models at package mxnet are migrated from mxnet symbols and mxnet model zoom.

Installation

Requirements:

  • python3
  • tensorflow >= 1.0 for package tensorflow
  • mxnet >= 0.10 for package mxnet
  • opencv3
  • yaml
  • numpy
  • matplotlib
  • collections
  • PIL(pillow)

Structure

  • mxnet
    • cv_tools (tools process image and video using opencv)
    • imagenet (trained weights on imagenet-1000 from mxnet model zoo, .json file and .params file)
    • model (model from mxnet model zoom,you can add new models to this package as you need)
    • tools (some useful tools,train log visualized tool for now)
    • trainedmodel (trained model on your own dataset)

Usage

train on your dataset

python train.py --parampath params.yaml

All the parameters needed for training configured in params.yaml.

test with your model

python batch_eval.py \
    --imagepath /home/gy/gitpro/mlAlgorithms/mxnet/cv_tools/image \
    --modelprefix trainedmodel/model \
    --epoch 2

label.txt is a file contains
classification index---classification name pair

process your dataset

At package cv_tools

ptrhon pack.py \
    --imagepath iamges \
    --datapath traindata \
    --imgsize "499,499" \
    --channel 3 \
    --slice 1

Todo

- [ ] process input
- [ ] add mxnet Rec format
- [ ] using gluon
- [ ] multiple gpus