API change/update

[ankitaggarwal011]

Currently, the API provides a way to give the location of the input file for processing and requires the desired location of the output. This inhibits the flexibility of the library to some extent.

Another way would require the API to deal directly with the image matrices rather than the location and the onus of I/O would fall on the developer. However, completely eliminating the current API would pose a problem to people who are looking for plug-n-play solution. Thus, it might be better to support both the forms of API.

Any suggestions are welcome. Thanks!

[sahildua2305]

@ankitaggarwal011 Do you intend to take input file blob from the user and then process it?

[ankitaggarwal011]

Hi @sahildua2305, I was thinking about including the api such that the I/O operations are taken care by the user, and the api will take an input image Numpy matrix and provide the processed result as a Numpy output image matrix.

One advantage of constraining the type of input as Numpy image matrix would be easy validation of input, however, it maybe wise to include the other types (blob, array) of the input image. A certain restriction has to be made on the type of input. What do you think about this?

Thanks for helping out. Appreciate it!

[bradstimpson]

Hi @ankitaggarwal011 I can help out here. Any objections if I jump in?

[ankitaggarwal011]

@bradstimpson, please go ahead. Your contributions are most welcome. Thanks.