from os import listdir, walk from os.path import isfile, isdir, join, splitext, exists, getmtime from random import seed, randint, choice import re import json import datetime import argparse parser = argparse.ArgumentParser(description='Generate BBBicycles split.') parser.add_argument('-p', '--path', type=str, required=True, help='directory containing the ID folders') args = parser.parse_args() path = args.path random_seed = seed(1337) train_val_bike_type_split = 10 img_regex = re.compile('(^img.\d*[.]png$)') dir_regex = re.compile('(^\w+_)') train = open("bike_train.txt", "w") query_v = open("bike_query_val.txt", "w") galley_v = open("bike_gallery_val.txt", "w") query_t = open("bike_query_test.txt", "w") galley_t = open("bike_gallery_test.txt", "w") num_ids = 0 num_imgs = 0 num_damaged_imgs = 0 num_broken_imgs = 0 num_bent_imgs = 0 num_missingpart_imgs = 0 nums_missingpart_imgs = [0 for i in range(5)] models_dist = {} num_train_ids = 0 num_train_imgs = 0 num_damaged_train_imgs = 0 num_broken_train_imgs = 0 num_bent_train_imgs = 0 num_missingpart_train_imgs = 0 nums_missingpart_train_imgs = [0 for i in range(5)] models_dist_train = {} num_val_ids = 0 num_val_imgs = 0 num_damaged_val_imgs = 0 num_broken_val_imgs = 0 num_bent_val_imgs = 0 num_missingpart_val_imgs = 0 nums_missingpart_val_imgs = [0 for i in range(5)] models_dist_val = {} num_test_ids = 0 num_test_imgs = 0 num_damaged_test_imgs = 0 num_broken_test_imgs = 0 num_bent_test_imgs = 0 num_missingpart_test_imgs = 0 nums_missingpart_test_imgs = [0 for i in range(5)] models_dist_test = {} for id, bike in enumerate(listdir(path)): if isdir(join(path, bike)) and dir_regex.match(bike) and exists(join(path, bike, "before")) and len(listdir(join(path, bike, "before"))) != 0 and exists(join(path, bike, "after")) and len(listdir(join(path, bike, "after"))) != 0 and exists(join(path, bike, "fixed_data.json")): #aggiungere: prendere json dell'identità per estrarre info json_fixed = open(join(path, bike, 'fixed_data.json'),) data_fixed = json.load(json_fixed) if str(data_fixed['Bike Type']) not in models_dist: models_dist[str(data_fixed['Bike Type'])] = {str(data_fixed['Model']): 1} elif str(data_fixed['Model']) not in models_dist[str(data_fixed['Bike Type'])]: models_dist[str(data_fixed['Bike Type'])][str(data_fixed['Model'])] = 1 else: models_dist[str(data_fixed['Bike Type'])][str(data_fixed['Model'])] = models_dist[str(data_fixed['Bike Type'])][str(data_fixed['Model'])] + 1 c=0 num_ids = num_ids + 1 for type in ["before", "after"]: for file in listdir(join(path, bike, type)): if img_regex.match(file): if exists(join(path, bike, type, splitext(file)[0] + '_variable.json')): json_var = open(join(path, bike, type, splitext(file)[0] + '_variable.json'),) data = json.load(json_var) dmgid = 0 if type == "before" else int(data["Damage Type"]) missid = "00000" if type == "before" else str(data["Removed Parts"]) json_var.close() if dmgid != 0: num_damaged_imgs = num_damaged_imgs + 1 if dmgid == 2 or dmgid == 3: num_broken_imgs = num_broken_imgs + 1 if dmgid == 1 or dmgid == 3: num_bent_imgs = num_bent_imgs + 1 if missid != "00000" : num_missingpart_imgs = num_missingpart_imgs + 1 for i in range(5): if missid[i] == "1": nums_missingpart_imgs[i] = nums_missingpart_imgs[i] + 1 num_imgs = num_imgs + 1 c=c+1 else: print(bike) if c != 14: print(bike) if str(data_fixed['Model']) in ['mfactory ', 'ghost', 'oldbike', 'rondo', 'verdona']: #Train if str(data_fixed['Bike Type']) not in models_dist_train: models_dist_train[str(data_fixed['Bike Type'])] = {str(data_fixed['Model']): 1} elif str(data_fixed['Model']) not in models_dist_train[str(data_fixed['Bike Type'])]: models_dist_train[str(data_fixed['Bike Type'])][str(data_fixed['Model'])] = 1 else: models_dist_train[str(data_fixed['Bike Type'])][str(data_fixed['Model'])] = models_dist_train[str(data_fixed['Bike Type'])][str(data_fixed['Model'])] + 1 num_train_ids = num_train_ids + 1 for type in ["before", "after"]: for file in listdir(join(path, bike, type)): if img_regex.match(file) and exists(join(path, bike, type, splitext(file)[0] + '_variable.json')): img_path = join(bike, type, file) json_var = open(join(path, bike, type, splitext(file)[0] + '_variable.json'),) data = json.load(json_var) camid = int(data["Focal Length"]) viewid = int(data["Viewing Side"]) dmgid = 0 if type == "before" else int(data["Damage Type"]) missid = "00000" if type == "before" else str(data["Removed Parts"]) train.write("{} {} {} {} {} {}\n".format(img_path, id, camid, viewid, dmgid, missid)) json_var.close() if dmgid != 0: num_damaged_train_imgs = num_damaged_train_imgs + 1 if dmgid == 2 or dmgid == 3: num_broken_train_imgs = num_broken_train_imgs + 1 if dmgid == 1 or dmgid == 3: num_bent_train_imgs = num_bent_train_imgs + 1 if missid != "00000" : num_missingpart_train_imgs = num_missingpart_train_imgs + 1 for i in range(5): if missid[i] == "1": nums_missingpart_train_imgs[i] = nums_missingpart_train_imgs[i] + 1 num_train_imgs = num_train_imgs + 1 else: if str(data_fixed['Model']) not in ['mirage', 'gbike', 'enduro']: if str(data_fixed['Model']) not in ['becane', 'btwin', 'croad'] and randint(0, 100) > train_val_bike_type_split: if str(data_fixed['Bike Type']) not in models_dist_train: models_dist_train[str(data_fixed['Bike Type'])] = {str(data_fixed['Model']): 1} elif str(data_fixed['Model']) not in models_dist_train[str(data_fixed['Bike Type'])]: models_dist_train[str(data_fixed['Bike Type'])][str(data_fixed['Model'])] = 1 else: models_dist_train[str(data_fixed['Bike Type'])][str(data_fixed['Model'])] = models_dist_train[str(data_fixed['Bike Type'])][str(data_fixed['Model'])] + 1 num_train_ids = num_train_ids + 1 for type in ["before", "after"]: for file in listdir(join(path, bike, type)): if img_regex.match(file) and exists(join(path, bike, type, splitext(file)[0] + '_variable.json')): img_path = join(bike, type, file) json_var = open(join(path, bike, type, splitext(file)[0] + '_variable.json'),) data = json.load(json_var) camid = int(data["Focal Length"]) viewid = int(data["Viewing Side"]) dmgid = 0 if type == "before" else int(data["Damage Type"]) missid = "00000" if type == "before" else str(data["Removed Parts"]) train.write("{} {} {} {} {} {}\n".format(img_path, id, camid, viewid, dmgid, missid)) json_var.close() if dmgid != 0: num_damaged_train_imgs = num_damaged_train_imgs + 1 if dmgid == 2 or dmgid == 3: num_broken_train_imgs = num_broken_train_imgs + 1 if dmgid == 1 or dmgid == 3: num_bent_train_imgs = num_bent_train_imgs + 1 if missid != "00000" : num_missingpart_train_imgs = num_missingpart_train_imgs + 1 for i in range(5): if missid[i] == "1": nums_missingpart_train_imgs[i] = nums_missingpart_train_imgs[i] + 1 num_train_imgs = num_train_imgs + 1 else: #Val if str(data_fixed['Bike Type']) not in models_dist_val: models_dist_val[str(data_fixed['Bike Type'])] = {str(data_fixed['Model']): 1} elif str(data_fixed['Model']) not in models_dist_val[str(data_fixed['Bike Type'])]: models_dist_val[str(data_fixed['Bike Type'])][str(data_fixed['Model'])] = 1 else: models_dist_val[str(data_fixed['Bike Type'])][str(data_fixed['Model'])] = models_dist_val[str(data_fixed['Bike Type'])][str(data_fixed['Model'])] + 1 num_val_ids = num_val_ids + 1 for type in ["before", "after"]: files = [f for f in listdir(join(path, bike, type)) if img_regex.match(f) and exists(join(path, bike, type, splitext(f)[0] + '_variable.json'))] file = choice(files) img_path = join(bike, type, file) json_var = open(join(path, bike, type, splitext(file)[0] + '_variable.json'),) data = json.load(json_var) camid = int(data["Focal Length"]) viewid = int(data["Viewing Side"]) dmgid = 0 if type == "before" else int(data["Damage Type"]) missid = "00000" if type == "before" else str(data["Removed Parts"]) if type == "before" : galley_v.write("{} {} {} {} {} {}\n".format(img_path, id, camid, viewid, dmgid, missid)) else: query_v.write("{} {} {} {} {} {}\n".format(img_path, id, camid, viewid, dmgid, missid)) json_var.close() if dmgid != 0: num_damaged_val_imgs = num_damaged_val_imgs + 1 if dmgid == 2 or dmgid == 3: num_broken_val_imgs = num_broken_val_imgs + 1 if dmgid == 1 or dmgid == 3: num_bent_val_imgs = num_bent_val_imgs + 1 if missid != "00000" : num_missingpart_val_imgs = num_missingpart_val_imgs + 1 for i in range(5): if missid[i] == "1": nums_missingpart_val_imgs[i] = nums_missingpart_val_imgs[i] + 1 num_val_imgs = num_val_imgs + 1 else: #Test if str(data_fixed['Bike Type']) not in models_dist_test: models_dist_test[str(data_fixed['Bike Type'])] = {str(data_fixed['Model']): 1} elif str(data_fixed['Model']) not in models_dist_test[str(data_fixed['Bike Type'])]: models_dist_test[str(data_fixed['Bike Type'])][str(data_fixed['Model'])] = 1 else: models_dist_test[str(data_fixed['Bike Type'])][str(data_fixed['Model'])] = models_dist_test[str(data_fixed['Bike Type'])][str(data_fixed['Model'])] + 1 num_test_ids = num_test_ids + 1 for type in ["before", "after"]: files = [f for f in listdir(join(path, bike, type)) if img_regex.match(f) and exists(join(path, bike, type, splitext(f)[0] + '_variable.json'))] if not files: print(join(path, bike, type)) file = choice(files) img_path = join(bike, type, file) json_var = open(join(path, bike, type, splitext(file)[0] + '_variable.json'),) data = json.load(json_var) camid = int(data["Focal Length"]) viewid = int(data["Viewing Side"]) dmgid = 0 if type == "before" else int(data["Damage Type"]) missid = "00000" if type == "before" else str(data["Removed Parts"]) if type == "before" : galley_t.write("{} {} {} {} {} {}\n".format(img_path, id, camid, viewid, dmgid, missid)) else: query_t.write("{} {} {} {} {} {}\n".format(img_path, id, camid, viewid, dmgid, missid)) json_var.close() if dmgid != 0: num_damaged_test_imgs = num_damaged_test_imgs + 1 if dmgid == 2 or dmgid == 3: num_broken_test_imgs = num_broken_test_imgs + 1 if dmgid == 1 or dmgid == 3: num_bent_test_imgs = num_bent_test_imgs + 1 if missid != "00000" : num_missingpart_test_imgs = num_missingpart_test_imgs + 1 for i in range(5): if missid[i] == "1": nums_missingpart_test_imgs[i] = nums_missingpart_test_imgs[i] + 1 num_test_imgs = num_test_imgs + 1 json_fixed.close() else: print(bike) train.close() query_v.close() galley_v.close() query_t.close() galley_t.close() data = {} data["General"] = [] data["General"].append({ 'Num IDs': num_ids, 'Num Bike types': len(models_dist.keys()), 'Num Models': sum(len(models_dist[k].keys()) for k in models_dist.keys()), 'Num images': num_imgs, 'Num bent images': num_bent_imgs, 'Num broken images': num_broken_imgs, 'Num damaged images': num_damaged_imgs, 'Num images with missing parts': num_missingpart_imgs, 'Num images with missing Front Wheel': nums_missingpart_imgs[0], 'Num images with missing Rear Wheel': nums_missingpart_imgs[1], 'Num images with missing Seat': nums_missingpart_imgs[2], 'Num images with missing Handlebar': nums_missingpart_imgs[3], 'Num images with missing Pedals': nums_missingpart_imgs[4] }) data["Train"] = [] data["Train"].append({ 'Num IDs': num_train_ids, 'Num Bike types': len(models_dist_train.keys()), 'Num Models': sum(len(models_dist_train[k].keys()) for k in models_dist_train.keys()), 'Num images': num_train_imgs, 'Num bent images': num_bent_train_imgs, 'Num broken images': num_broken_train_imgs, 'Num damaged images': num_damaged_train_imgs, 'Num images with missing parts': num_missingpart_train_imgs, 'Num images with missing Front Wheel': nums_missingpart_train_imgs[0], 'Num images with missing Rear Wheel': nums_missingpart_train_imgs[1], 'Num images with missing Seat': nums_missingpart_train_imgs[2], 'Num images with missing Handlebar': nums_missingpart_train_imgs[3], 'Num images with missing Pedals': nums_missingpart_train_imgs[4] }) data["Validation"] = [] data["Validation"].append({ 'Num IDs': num_val_ids, 'Num Bike types': len(models_dist_val.keys()), 'Num Models': sum(len(models_dist_val[k].keys()) for k in models_dist_val.keys()), 'Num images': num_val_imgs, 'Num bent images': num_bent_val_imgs, 'Num broken images': num_broken_val_imgs, 'Num damaged images': num_damaged_val_imgs, 'Num images with missing parts': num_missingpart_val_imgs, 'Num images with missing Front Wheel': nums_missingpart_val_imgs[0], 'Num images with missing Rear Wheel': nums_missingpart_val_imgs[1], 'Num images with missing Seat': nums_missingpart_val_imgs[2], 'Num images with missing Handlebar': nums_missingpart_val_imgs[3], 'Num images with missing Pedals': nums_missingpart_val_imgs[4] }) data["Test"] = [] data["Test"].append({ 'Num IDs': num_test_ids, 'Num Bike types': len(models_dist_test.keys()), 'Num Models': sum(len(models_dist_test[k].keys()) for k in models_dist_test.keys()), 'Num images': num_test_imgs, 'Num bent images': num_bent_test_imgs, 'Num broken images': num_broken_test_imgs, 'Num damaged images': num_damaged_test_imgs, 'Num images with missing parts': num_missingpart_test_imgs, 'Num images with missing Front Wheel': nums_missingpart_test_imgs[0], 'Num images with missing Rear Wheel': nums_missingpart_test_imgs[1], 'Num images with missing Seat': nums_missingpart_test_imgs[2], 'Num images with missing Handlebar': nums_missingpart_test_imgs[3], 'Num images with missing Pedals': nums_missingpart_test_imgs[4] }) with open('bike_current_split_stats.json', 'w', encoding='utf-8') as f: json.dump(data, f, ensure_ascii=False, indent=4)