o Ni<@s~dZddlmZddlmZddlmZddlZddlZddlm m Z ddl m ZdZdZdZd ZGd d d ejjZdS) z300W-LP Dataset.)absolute_import)division)print_functionNzKhttps://drive.google.com/uc?export=download&id=0B7OEHD3T4eCkVGs0TkhUWFN6N1kzAhttp://www.cbsr.ia.ac.cn/users/xiangyuzhu/projects/3DDFA/main.htman@article{DBLP:journals/corr/ZhuLLSL15, author = {Xiangyu Zhu and Zhen Lei and Xiaoming Liu and Hailin Shi and Stan Z. Li}, title = {Face Alignment Across Large Poses: {A} 3D Solution}, journal = {CoRR}, volume = {abs/1511.07212}, year = {2015}, url = {http://arxiv.org/abs/1511.07212}, archivePrefix = {arXiv}, eprint = {1511.07212}, timestamp = {Mon, 13 Aug 2018 16:48:23 +0200}, biburl = {https://dblp.org/rec/bib/journals/corr/ZhuLLSL15}, bibsource = {dblp computer science bibliography, https://dblp.org} } a300W-LP Dataset is expanded from 300W, which standardises multiple alignment databases with 68 landmarks, including AFW, LFPW, HELEN, IBUG and XM2VTS. With 300W, 300W-LP adopt the proposed face profiling to generate 61,225 samples across large poses (1,786 from IBUG, 5,207 from AFW, 16,556 from LFPW and 37,676 from HELEN, XM2VTS is not used). The dataset can be employed as the training set for the following computer vision tasks: face attribute recognition and landmark (or facial part) locaization. c@s4eZdZdZejdZddZddZ ddZ d S) The300wLpz300W-LP dataset.z1.0.0cCstjj|ttjtjjdddtjjdtj dtjjdtj dtjjdtj dtjjdtj dtjjdtj dtjjdtj dtjjd tj dtjjd tj dtjjd tj dtjjdtj dd t t d S) N)rjpeg)shapeencoding_format)D)r dtype)) ))) imagelandmarks_origin landmarks_2d landmarks_3droi illum_params color_params tex_params shape_params exp_params pose_params)builder descriptionfeatureshomepagecitation) tfdscore DatasetInfo _DESCRIPTIONr! FeaturesDictImageTensortffloat32 _PROJECT_URL _CITATION)selfr0X/home/ubuntu/.local/lib/python3.10/site-packages/tensorflow_datasets/image/the300w_lp.py_infoFs*zThe300wLp._infocCs.|t}tjjtjjdtj |didgS)zReturns SplitGenerators.image_dir_path300W_LP)name gen_kwargs) download_and_extract _DATASET_URLr$r%SplitGeneratorSplitTRAINospathjoin)r/ dl_managerextracted_pathr0r0r1_split_generatorsgs zThe300wLp._split_generatorsc cstjjjtj|dd}dd|D}dd|D}t|||D]\}}}tjj|d}t j j j j |} Wdn1sCwY| dj} | d tj} | d d tj} | d d gtj} | dd gtj} | dd gtj}| dd gtj}| dd gtj}| dd gtj}tjj|d%}t j j j j |}|dd tj}|dd tj}Wdn1swY|| ||| | | ||||d }tj||fVq"dS)zYields examples.z[!Code]*[!_Flip]/[!_]*.jpg)patterncSsg|]}|ddqS)jpgmatreplace.0sr0r0r1 vsz0The300wLp._generate_examples..cSs g|] }|ddddqS)r4z300W_LP/landmarksz.jpgz_pts.matrErGr0r0r1rJwsrbNpt2dg |@rr Illum_Para Color_ParaTex_Para Shape_ParaExp_Para Pose_Parapts_2dpts_3dr)r+iogfileglobr<r=r>zipGFiler$r% lazy_importsscipyloadmatTastypenpr,reshapebasename)r/r3 image_files label_fileslandmark_files image_file label_file landmark_filefrD pt2d_originrrrrrrrldm_matrLpt3drecordr0r0r1_generate_examplesrsT   zThe300wLp._generate_examplesN) __name__ __module__ __qualname____doc__r$r%VersionVERSIONr2rArnr0r0r0r1rAs  ! r)rr __future__rrrr<numpyr`tensorflow.compat.v2compatv2r+tensorflow_datasets.public_api public_apir$r8r-r.r'r%GeneratorBasedBuilderrr0r0r0r1s