def initialize(self, opt): self.opt = opt self.root = opt.dataroot self.diction = {} # input A (label maps) dir_A = '_A' if self.opt.label_nc == 0 else '_label' self.dir_A = os.path.join(opt.dataroot, opt.phase + dir_A) self.A_paths = sorted(make_dataset(self.dir_A)) self.fine_height = 256 self.fine_width = 192 self.radius = 5 # input A test (label maps) dir_A = '_A' if self.opt.label_nc == 0 else '_label' self.dir_A = os.path.join(opt.dataroot, opt.phase + dir_A) self.A_paths = sorted(make_dataset_test(self.dir_A)) # input B (real images) dir_B = '_B' if self.opt.label_nc == 0 else '_img' self.dir_B = os.path.join(opt.dataroot, opt.phase + dir_B) self.B_paths = sorted(make_dataset(self.dir_B)) self.dataset_size = len(self.A_paths) self.build_index(self.B_paths) dir_E = '_edge' self.dir_E = os.path.join(opt.dataroot, opt.phase + dir_E) self.E_paths = sorted(make_dataset(self.dir_E)) self.ER_paths = make_dataset(self.dir_E) dir_M = '_mask' self.dir_M = os.path.join(opt.dataroot, opt.phase + dir_M) self.M_paths = sorted(make_dataset(self.dir_M)) self.MR_paths = make_dataset(self.dir_M) dir_MC = '_colormask' self.dir_MC = os.path.join(opt.dataroot, opt.phase + dir_MC) self.MC_paths = sorted(make_dataset(self.dir_MC)) self.MCR_paths = make_dataset(self.dir_MC) dir_C = '_color' self.dir_C = os.path.join(opt.dataroot, opt.phase + dir_C) self.C_paths = sorted(make_dataset(self.dir_C)) self.CR_paths = make_dataset(self.dir_C) # self.build_index(self.C_paths) dir_A = '_A' if self.opt.label_nc == 0 else '_label' self.dir_A = os.path.join(opt.dataroot, opt.phase + dir_A) self.A_paths = sorted(make_dataset_test(self.dir_A))
def initialize(self, opt): self.opt = opt self.root = opt.dataroot print("label_nc: ", self.opt.label_nc) ### input A (label maps) if opt.isTrain or opt.use_encoded_image: dir_A = '_A' if self.opt.label_nc == 0 else '_label' self.dir_A = os.path.join(opt.dataroot, opt.phase + dir_A) self.A_paths = sorted(make_dataset(self.dir_A)) self.AR_paths = make_dataset(self.dir_A) ### input A inter 1 (label maps) if opt.isTrain or opt.use_encoded_image: dir_A_inter_1 = '_label_inter_1' self.dir_A_inter_1 = os.path.join(opt.dataroot, opt.phase + dir_A_inter_1) self.A_paths_inter_1 = sorted(make_dataset(self.dir_A_inter_1)) ### input A inter 2 (label maps) if opt.isTrain or opt.use_encoded_image: dir_A_inter_2 = '_label_inter_2' self.dir_A_inter_2 = os.path.join(opt.dataroot, opt.phase + dir_A_inter_2) self.A_paths_inter_2 = sorted(make_dataset(self.dir_A_inter_2)) ### input A test (label maps) if not (opt.isTrain or opt.use_encoded_image): dir_A = '_A' if self.opt.label_nc == 0 else '_label' self.dir_A = os.path.join(opt.dataroot, opt.phase + dir_A) self.A_paths = sorted(make_dataset_test(self.dir_A)) dir_AR = '_AR' if self.opt.label_nc == 0 else '_labelref' self.dir_AR = os.path.join(opt.dataroot, opt.phase + dir_AR) self.AR_paths = sorted(make_dataset_test(self.dir_AR)) ### input B (real images) dir_B = '_B' if self.opt.label_nc == 0 else '_img' self.dir_B = os.path.join(opt.dataroot, opt.phase + dir_B) self.B_paths = sorted(make_dataset(self.dir_B)) self.BR_paths = sorted(make_dataset(self.dir_B)) self.dataset_size = len(self.A_paths)
def initialize(self, opt): self.opt = opt self.root = opt.dataroot ### input A (label maps) if opt.isTrain or opt.use_encoded_image: dir_A = '_A' if self.opt.label_nc == 0 else '_label' self.dir_A = os.path.join(opt.dataroot, opt.phase + dir_A) self.A_paths = sorted(make_dataset(self.dir_A)) ### input A test (label maps) if not (opt.isTrain or opt.use_encoded_image): dir_A = '_A' if self.opt.label_nc == 0 else '_label' self.dir_A = os.path.join(opt.dataroot, opt.phase + dir_A) self.A_paths = sorted(make_dataset_test(self.dir_A)) ### input B (real images) dir_B = '_B' if self.opt.label_nc == 0 else '_img' self.dir_B = os.path.join(opt.dataroot, opt.phase + dir_B) self.B_paths = sorted(make_dataset(self.dir_B)) self.BR_paths = sorted(make_dataset(self.dir_B)) self.dataset_size = len(self.A_paths)
def initialize(self, opt): self.opt = opt self.root = opt.dataroot self.diction = {} self.fine_height = 256 self.fine_width = 192 self.radius = 5 # load data list from pairs file human_names = [] cloth_names = [] with open(os.path.join(opt.dataroot, opt.datapairs), 'r') as f: for line in f.readlines(): h_name, c_name = line.strip().split() human_names.append(h_name) cloth_names.append(c_name) self.human_names = human_names self.cloth_names = cloth_names self.dataset_size = len(human_names) # input A (label maps) dir_A = '_A' if self.opt.label_nc == 0 else '_label' self.dir_A = os.path.join(opt.dataroot, opt.phase + dir_A) self.A_paths = sorted(make_dataset(self.dir_A)) self.fine_height = 256 self.fine_width = 192 self.radius = 5 # input A test (label maps) dir_A = '_A' if self.opt.label_nc == 0 else '_label' self.dir_A = os.path.join(opt.dataroot, opt.phase + dir_A) self.A_paths = sorted(make_dataset_test(self.dir_A)) # input B (real images) dir_B = '_B' if self.opt.label_nc == 0 else '_img' self.dir_B = os.path.join(opt.dataroot, opt.phase + dir_B) self.B_paths = sorted(make_dataset(self.dir_B)) self.dataset_size = len(self.A_paths) self.build_index(self.B_paths) dir_E = '_edge' self.dir_E = os.path.join(opt.dataroot, opt.phase + dir_E) self.E_paths = sorted(make_dataset(self.dir_E)) self.ER_paths = make_dataset(self.dir_E) dir_M = '_mask' self.dir_M = os.path.join(opt.dataroot, opt.phase + dir_M) self.M_paths = sorted(make_dataset(self.dir_M)) self.MR_paths = make_dataset(self.dir_M) dir_MC = '_colormask' self.dir_MC = os.path.join(opt.dataroot, opt.phase + dir_MC) self.MC_paths = sorted(make_dataset(self.dir_MC)) self.MCR_paths = make_dataset(self.dir_MC) dir_C = '_color' self.dir_C = os.path.join(opt.dataroot, opt.phase + dir_C) self.C_paths = sorted(make_dataset(self.dir_C)) self.CR_paths = make_dataset(self.dir_C) # self.build_index(self.C_paths) dir_A = '_A' if self.opt.label_nc == 0 else '_label' self.dir_A = os.path.join(opt.dataroot, opt.phase + dir_A) self.A_paths = sorted(make_dataset_test(self.dir_A))
def initialize(self, opt): self.opt = opt self.root = opt.dataroot self.diction = {} im_names = [] c_names = [] if opt.data_list != None: opt.phase = "combined" with open(osp.join(opt.data_list), 'r') as f: for line in f.readlines(): im_name, c_name = line.strip().split() im_names.append(im_name) c_names.append(c_name) ### input A (label maps) if opt.isTrain or opt.use_encoded_image: dir_A = '_A' if self.opt.label_nc == 0 else '_label' self.dir_A = os.path.join(opt.dataroot, opt.phase + dir_A) self.A_paths = sorted(make_dataset(self.dir_A)) self.AR_paths = make_dataset(self.dir_A) if opt.data_list != None: self.A_paths = [ os.path.join(self.dir_A, n).replace(".jpg", ".png") for n in im_names ] self.fine_height = 256 self.fine_width = 192 self.radius = 5 ### input A test (label maps) if not (opt.isTrain or opt.use_encoded_image): dir_A = '_A' if self.opt.label_nc == 0 else '_label' self.dir_A = os.path.join(opt.dataroot, opt.phase + dir_A) self.A_paths = sorted(make_dataset_test(self.dir_A)) dir_AR = '_AR' if self.opt.label_nc == 0 else '_labelref' self.dir_AR = os.path.join(opt.dataroot, opt.phase + dir_AR) self.AR_paths = sorted(make_dataset_test(self.dir_AR)) ### input B (real images) dir_B = '_B' if self.opt.label_nc == 0 else '_img' self.dir_B = os.path.join(opt.dataroot, opt.phase + dir_B) self.B_paths = sorted(make_dataset(self.dir_B)) self.BR_paths = sorted(make_dataset(self.dir_B)) if opt.data_list != None: self.B_paths = [os.path.join(self.dir_B, n) for n in im_names] self.dataset_size = len(self.A_paths) self.build_index(self.B_paths) ### input E (edge_maps) if opt.isTrain or opt.use_encoded_image: dir_E = '_edge' self.dir_E = os.path.join(opt.dataroot, opt.phase + dir_E) self.E_paths = sorted(make_dataset(self.dir_E)) self.ER_paths = make_dataset(self.dir_E) if opt.data_list != None: self.E_paths = [os.path.join(self.dir_E, n) for n in c_names] # ### input M (masks) # if opt.isTrain or opt.use_encoded_image: # dir_M = '_mask' # self.dir_M = os.path.join(opt.dataroot, opt.phase + dir_M) # self.M_paths = sorted(make_dataset(self.dir_M)) # self.MR_paths = make_dataset(self.dir_M) # # ### input MC(color_masks) # if opt.isTrain or opt.use_encoded_image: # dir_MC = '_colormask' # self.dir_MC = os.path.join(opt.dataroot, opt.phase + dir_MC) # self.MC_paths = sorted(make_dataset(self.dir_MC)) # self.MCR_paths = make_dataset(self.dir_MC) ### input C(color) if opt.isTrain or opt.use_encoded_image: dir_C = '_color' self.dir_C = os.path.join(opt.dataroot, opt.phase + dir_C) self.C_paths = sorted(make_dataset(self.dir_C)) self.CR_paths = make_dataset(self.dir_C) if opt.data_list != None: self.C_paths = [os.path.join(self.dir_C, n) for n in c_names] # self.build_index(self.C_paths) ### input A test (label maps) if not (opt.isTrain or opt.use_encoded_image): dir_A = '_A' if self.opt.label_nc == 0 else '_label' self.dir_A = os.path.join(opt.dataroot, opt.phase + dir_A) self.A_paths = sorted(make_dataset_test(self.dir_A))
def initialize(self, opt): self.opt = opt self.root = opt.dataroot self.diction = {} ### input A (label maps) if opt.isTrain or opt.use_encoded_image: dir_A = '_A' if self.opt.label_nc == 0 else '_label' self.dir_A = os.path.join(opt.dataroot, opt.phase + dir_A) self.A_paths = sorted(make_dataset(self.dir_A)) self.AR_paths = make_dataset(self.dir_A) self.fine_height = 256 self.fine_width = 192 self.radius = 5 ### input A test (label maps) if not (opt.isTrain or opt.use_encoded_image): dir_A = '_A' if self.opt.label_nc == 0 else '_label' self.dir_A = os.path.join(opt.dataroot, opt.phase + dir_A) self.A_paths = sorted(make_dataset_test(self.dir_A)) dir_AR = '_AR' if self.opt.label_nc == 0 else '_labelref' self.dir_AR = os.path.join(opt.dataroot, opt.phase + dir_AR) self.AR_paths = sorted(make_dataset_test(self.dir_AR)) ### input B (real images) dir_B = '_B' if self.opt.label_nc == 0 else '_img' self.dir_B = os.path.join(opt.dataroot, opt.phase + dir_B) self.B_paths = sorted(make_dataset(self.dir_B)) self.BR_paths = sorted(make_dataset(self.dir_B)) self.dataset_size = len(self.A_paths) self.build_index(self.B_paths) ### input E (edge_maps) if opt.isTrain or opt.use_encoded_image: dir_E = '_edge' self.dir_E = os.path.join(opt.dataroot, opt.phase + dir_E) self.E_paths = sorted(make_dataset(self.dir_E)) self.ER_paths = make_dataset(self.dir_E) ### input M (masks) if opt.isTrain or opt.use_encoded_image: dir_M = '_mask' self.dir_M = os.path.join(opt.dataroot, opt.phase + dir_M) self.M_paths = sorted(make_dataset(self.dir_M)) self.MR_paths = make_dataset(self.dir_M) ### input MC(color_masks) if opt.isTrain or opt.use_encoded_image: dir_MC = '_colormask' self.dir_MC = os.path.join(opt.dataroot, opt.phase + dir_MC) self.MC_paths = sorted(make_dataset(self.dir_MC)) self.MCR_paths = make_dataset(self.dir_MC) ### input C(color) if opt.isTrain or opt.use_encoded_image: dir_C = '_color' self.dir_C = os.path.join(opt.dataroot, opt.phase + dir_C) self.C_paths = sorted(make_dataset(self.dir_C)) self.CR_paths = make_dataset(self.dir_C) # self.build_index(self.C_paths) ### input A test (label maps) if not (opt.isTrain or opt.use_encoded_image): dir_A = '_A' if self.opt.label_nc == 0 else '_label' self.dir_A = os.path.join(opt.dataroot, opt.phase + dir_A) self.A_paths = sorted(make_dataset_test(self.dir_A))
def initialize(self, opt): self.opt = opt self.root = opt.dataroot self.diction = {} ### input A (label maps) if opt.isTrain or opt.use_encoded_image: dir_A = '_A' if self.opt.label_nc == 0 else '_label' self.dir_A = os.path.join(opt.dataroot, opt.phase + dir_A) self.A_paths = sorted(make_dataset(self.dir_A)) self.AR_paths = make_dataset(self.dir_A) self.fine_height = 256 self.fine_width = 192 self.radius = 5 ### input A test (label maps) if not (opt.isTrain or opt.use_encoded_image): dir_A = '_A' if self.opt.label_nc == 0 else '_label' self.dir_A = os.path.join(opt.dataroot, opt.phase + dir_A) self.A_paths = sorted(make_dataset_test(self.dir_A)) dir_AR = '_AR' if self.opt.label_nc == 0 else '_labelref' self.dir_AR = os.path.join(opt.dataroot, opt.phase + dir_AR) self.AR_paths = sorted(make_dataset_test(self.dir_AR)) ### input B (real images) dir_B = '_B' if self.opt.label_nc == 0 else '_img' self.dir_B = os.path.join(opt.dataroot, opt.phase + dir_B) self.B_paths = sorted(make_dataset(self.dir_B)) self.BR_paths = sorted(make_dataset(self.dir_B)) self.dataset_size = len(self.A_paths) self.build_index(self.B_paths) ### input E (edge_maps) cloth if opt.isTrain or opt.use_encoded_image: dir_E = '_edge' self.dir_E = os.path.join(opt.dataroot, opt.phase + dir_E) #self.E_paths = sorted(make_dataset(self.dir_E)) self.E_paths = make_dataset_cloth(self.dir_E) self.ER_paths = make_dataset_cloth(self.dir_E) ### input M (masks) if opt.isTrain or opt.use_encoded_image: dir_M = '_mask' self.dir_M = os.path.join(opt.dataroot, opt.phase + dir_M) self.M_paths = sorted(make_dataset(self.dir_M)) self.MR_paths = make_dataset(self.dir_M) ### input MC(color_masks) cloth if opt.isTrain or opt.use_encoded_image: dir_MC = '_colormask' self.dir_MC = os.path.join(opt.dataroot, opt.phase + dir_MC) #self.MC_paths = sorted(make_dataset(self.dir_MC)) self.MC_paths = make_dataset_cloth(self.dir_MC) self.MCR_paths = make_dataset_cloth(self.dir_MC) ### input C(color) cloth if opt.isTrain or opt.use_encoded_image: dir_C = '_color' self.dir_C = os.path.join(opt.dataroot, opt.phase + dir_C) #self.C_paths = sorted(make_dataset(self.dir_C)) self.C_paths = make_dataset_cloth(self.dir_C) self.CR_paths = make_dataset_cloth(self.dir_C) # self.build_index(self.C_paths) ### input A test (label maps) if not (opt.isTrain or opt.use_encoded_image): dir_A = '_A' if self.opt.label_nc == 0 else '_label' self.dir_A = os.path.join(opt.dataroot, opt.phase + dir_A) self.A_paths = sorted(make_dataset_test(self.dir_A)) ### input VS (VTON Segmentation) if opt.isTrain or opt.use_encoded_image: dir_VS = '_seg' self.dir_VS = os.path.join(opt.dataroot, opt.phase + dir_VS) self.VS_paths = sorted(make_dataset(self.dir_VS)) self.VSR_paths = make_dataset(self.dir_VS) ### input S (Mesh Shape) if opt.isTrain or opt.use_encoded_image: dir_S = '_mesh' self.dir_S = os.path.join(opt.dataroot, opt.phase + dir_S) self.S_paths = sorted(make_dataset(self.dir_S)) self.SR_paths = make_dataset(self.dir_S) ### input D (Dense Pose) if opt.isTrain or opt.use_encoded_image: dir_D = '_dense' self.dir_D = os.path.join(opt.dataroot, opt.phase + dir_D) self.D_paths = sorted(make_dataset(self.dir_D)) self.DR_paths = make_dataset(self.dir_D) ### input CLM (Cloth Fashion Landmarks) if opt.isTrain or opt.use_encoded_image: dir_CLM = '_landmarks_cloth' self.dir_CLM = os.path.join(opt.dataroot, opt.phase + dir_CLM) self.CLM_paths = sorted(make_dataset(self.dir_CLM)) self.CLMR_paths = make_dataset(self.dir_CLM)