poda 0.4.15

A package to wrap tensor operation

README

NEWS

Date	News	Version
Mei 2019	poda is first update library	>=v0.1.0
Mei 2019	poda is second update library	>=v0.1.1
Mei 2019	poda is third update library	>=v0.1.2
June 2019	poda is fourth update library	>=v0.1.3
June 2019	poda is sixth update library	>=v0.1.4
August 2019	poda is second version and add Object detection	>=v0.2.0
August 2019	Adding segmentation model	>=v0.2.1
September 2019	Fixing tensor shape and rewrite parameters name	>=v0.2.2

Tensorflow Compatibility

Tensorflow version	Poda Version
1.14.0	>=v0.2.2

ABOUT PROJECT

This is a package to simplify tensorflow operations

DEPENDENCIES

1. Tensorflow (1.14.0)

For installing tensorflow, with GPU:

# python3 
pip3 install tensorflow-gpu
# python2
pip2 install tensorflow-gpu

Without GPU:

# python3 
pip3 install tensorflow
# python2
pip2 install tensorflow

HOW TO USE

Installing The Package

python setup.py install

or

pip3 install poda

Import The Package

• Tensor Operations

import tensorflow as tf
from poda.layers import *

• Convert Model to Tensorflow Serving

import tensorflow as tf
from poda.utils.convert_model_for_tensorflow_serving import *

• Transfer Learning Package

import tensorflow as tf
from poda.transfer_learning.Vgg16 import *

DOCKER

We already prepared the all in one docker for computer vision and deep learning libraries, including tensorflow 1.12, Opencv3.4.2 and contrib, CUDA 9, CUDNN 7, Keras, jupyter, numpy, sklearn, scipy, statsmodel, pandas, matplotlib, seaborn, flask, gunicorn etc. See the list of dockerfile below:

Docker: Ubuntu 16.04 with GPU (Cuda 9, cudnn 7.2) [TESTED]

• https://github.com/gideonmanurung/docker_utils/tree/master/docker_16.04

Docker: Ubuntu 18.04 with GPU (Cuda 9, cudnn 7.2)

• https://github.com/gideonmanurung/docker_utils/tree/master/docker_18.04

Docker: Ubuntu 16.04 without GPU (Cuda 9, cudnn 7.2) [TESTED]

• https://github.com/gideonmanurung/docker_utils/tree/without_gpu/docker_16.04

Docker: Ubuntu 18.04 without GPU (Cuda 9, cudnn 7.2) [TESTED]

• https://github.com/gideonmanurung/docker_utils/tree/without_gpu/docker_18.04