python - image comparison module - for machine learning purpose

# background to write 2 class 1) to compare 2 image similarity and output result as array of % 2) to create statistic report output as dictionary > Json / XML # requirement 1) write in python 2) use common library if possible 3) request opinion: can this logic run in pyspark / tensorflow or some other machine learning technology / GPU when image are "too many" # pseudo code import img_recognize img_recognize.main_chart( img_A ) img_recognize.sub_charts( [img_B, C, D, E, F, G ...] ) img_recognize.find_similar() output = [login to view URL]() print(output) { main_img : img_A, sub_imgs : [ img_b : 90%, B : 80%, C: 72%, D: 60%, E: 34% ] } import analysis_result conclusion = analysis_result( output ) print(conclusion) { statistic_report of the image comparison }