Coding Project: MNIST digits sequence generator

The goal of this project is to write a program that can generate images representing sequences of numbers, for data augmentation purposes. These images would be used to train classifiers and generative deep learning models. A script that saves examples of generated images is helpful in inspecting the characteristics of the generated images and in inspecting the trained models behaviours.

Specifications

• 0.digit dataset:MNIST database,
• 1.generate an image of a sequence
• 2.the digits have to be stacked horizontally
• 3.the spacing between them should follow a uniform distribution over a range determined by two user specified numbers. Each digit in the generated sequence is then chosen randomly from one of its representations in the MNIST dataset.
• 4.The width of the output image in pixels is specified by the user
• 5.the height should be 28 pixels
• 6.function:generate_numbers_sequence(digits, spacing_range, image_width):
• The generated image is saved in the current directory as a .png.

Note that besides implementing the expected function all, you are free to

implement as you wish. If you have time, you may also want to expand the project and think about additional methods for data expansion (warping, etc.).

* add the function of scale transformation 
* wait for expansion: rotation , shear, perspective and so on.

What should be provided

• a python module (or package) that defines the main function.
• That package should be written such as it can be imported by a 3rd party program. implementation details.
• you can import the function:[from imgAug import generate_numbers_sequence,random_scale_img]
• We expect the code to be organized, designed, tested and documented as if it were going into production.

excute result is outputed into the local folder:[dataout]'

• You can send the work assignement as a zipfile, tarball, etc. or better yet the whole code repository (git, hg, etc.). Just make sure it is not put in a public place !

zip , git

Scoring

• We value quality over feature-completeness. It is fine to leave things aside provided you call them out in your project's README. Part of the goal of this exercise is to help us identify what you consider production-ready code.

assess code:

• clarity: is the code organized in well defined functions, with separated concerns ? Is it implemented in a way that makes it difficult or simple to extend ? Depending on your advertised python expertise, we will also look whether the code is idiomatic python.
• documentation: does the README clearly and concisely explains the problem and solution ? Are technical tradeoffs explained ?
• testing and correctness: the code needs to do what is asked, and you need to be able to explain why it is correct. Ideally, the code would have some tests. We're not looking for full coverage (given time constraint) but just trying to get a feel for your testing skills.
• technical choices: do choices of libraries, architecture etc. seem appropriate for the chosen application?
• going the last mile: if you find the problem trivial, can you think of more advanced data augmentation techniques for the problem of handwritten digit recognition ? Can you think of a way to generate many images

dvanced data augmentation techniques
1.rotation  
2.shear
3.perspective 
4.and so on.

Can you think of a way to generate many images
the following function will output many images by user defination
:args.image_count
def main():
   os.makedirs(args.dataout_dir,exist_ok=True)
   df_train, gp = load_dataset(args.dataset_dir)
   for i in range(args.image_count):
       # img=generate_numbers_sequence(args.digits, args.spacing_range, args.image_width)
       img=generate_numbers_sequence_speedup(args.digits, args.spacing_range, args.image_width, df_train, gp)
       fn_id=datetime.datetime.now().strftime("%Y%m%d%H%S") + "_{0:02d}".format(i)
       fn_out=args.dataout_dir+"/"+fn_id+".png"
       cv2.imwrite(fn_out,img*255)

excute program

digit dataset

 download data from [MNIST database](http://yann.lecun.com/exdb/mnist/)
 save the data to [dataset]

package install

install the following package with pip or conda
$ opencv-python            4.1.1.26
$ pandas                   0.25.1
$ numpy                    1.16.4

function excute and explain

1.  generate_numbers_sequence(digits, spacing_range, image_width)
    generate one image of numbers sequence
    digits:[3, 5, 0]
    spacing_range:(10,30)
    image_width:100
2.  generate_numbers_sequence_speedup(digits, spacing_range, image_width,df_dt ,dt_gp, scaleFlg=True):
    digits:[3, 5, 0]
    spacing_range:(10,30)
    image_width:100
    scaleFlg:define the scale change or not
    df_dt ,dt_gp: prepare data by call function[load_dataset]
 * before excute this function , please firstly excute the following code,so when you produce multiple image,
 the speed will be up since the input data loading and preprocess is excute only one time.
  df_dt ,dt_gp = load_dataset(args.dataset_dir)
 * scaleFlg is used to define the scale transformation, if the true is defined, the scale transform will be 
   excuted , and the data augment become more strong
3.  random_scale_img(img, scale_in=0.2)
  scale transform function
  img:input image
  scale_in: define the scale, for example 0.2 is defined ,the output is random transform at the range [-0.2,0.2]
  for the axis of x , y

exctute sample

python imgAug.py \
--dataset_dir="dataset" \
--digits="[1,2,3]" \
--spacing_range="[5,10]" \
--image_width=200 \
--dataout_dir="testout50" \
--image_count=50