o Ni@sdZddlmZddlmZddlmZddlZddlmZddl m m Z ddl mZdZdZd Zd ZGd d d ejjZd dZddZdS)zSmallnorb dataset.)absolute_import)division)print_functionN)movesac\ @article{LeCun2004LearningMF, title={Learning methods for generic object recognition with invariance to pose and lighting}, author={Yann LeCun and Fu Jie Huang and L{\'e}on Bottou}, journal={Proceedings of the 2004 IEEE Computer Society Conference on Computer Vision and Pattern Recognition}, year={2004}, volume={2}, pages={II-104 Vol.2} } zghttps://cs.nyu.edu/~ylclab/data/norb-v1.0-small/smallnorb-5x46789x9x18x6x2x96x96-training-{type}.mat.gzzfhttps://cs.nyu.edu/~ylclab/data/norb-v1.0-small/smallnorb-5x01235x9x18x6x2x96x96-testing-{type}.mat.gza\ This database is intended for experiments in 3D object recognition from shape. It contains images of 50 toys belonging to 5 generic categories: four-legged animals, human figures, airplanes, trucks, and cars. The objects were imaged by two cameras under 6 lighting conditions, 9 elevations (30 to 70 degrees every 5 degrees), and 18 azimuths (0 to 340 every 20 degrees). The training set is composed of 5 instances of each category (instances 4, 6, 7, 8 and 9), and the test set of the remaining 5 instances (instances 0, 1, 2, 3, and 5). c@sDeZdZdZejddZejdgZddZ ddZ d d Z d S) SmallnorbzSmallnorb data set.2.0.0z6New split API (https://tensorflow.org/datasets/splits)z2.1.0c Cstjjddtjjddtjjgddtjjddtjjddtjjddtjjd dd }|jd kr;tj|d <tjj|ttj |d t ddS)N)`r)shape)zfour-legged animalsz human figures airplanestruckscars)names ) num_classes imageimage2label_categoryinstancelabel_elevation label_azimuthlabel_lightingridz0https://cs.nyu.edu/~ylclab/data/norb-v1.0-small/)rr)builder descriptionfeatureshomepagecitationsupervised_keys) tfdsrImage ClassLabelversionTextcore DatasetInfo _DESCRIPTION FeaturesDict _CITATION)self features_dictr/f/home/ubuntu/.local/lib/python3.10/site-packages/tensorflow_datasets/image_classification/smallnorb.py_info?s$        zSmallnorb._infoc Cstjddtjddtjddtjddtjddtjddd}||}tjjtjjt d|d|d|d d d tjjtjj t d |d |d|dd d gS)zReturns splits.dat)typecatinfo) training_dat training_cat training_info testing_dat testing_cat testing_infotrain_r6r7r8) split_prefixdat_pathcat_path info_path)name gen_kwargstest_r9r:r;) _TRAINING_URL_TEMPLATEformat_TESTING_URL_TEMPLATEdownload_and_extractr#r(SplitGeneratorSplitTRAINdictTEST)r- dl_manager filenamesfilesr/r/r0_split_generatorsas6       zSmallnorb._split_generatorsc cst|||\}}}tt|||D]/\}\} } } | d| d| | d| d| d| dd} |jdkr=d||| d<|| fVqd S) aGGenerate examples for the Smallnorb dataset. Args: split_prefix: Prefix that identifies the split. dat_path: Path to dat file of the chunk. cat_path: Path to cat file of the chunk. info_path: Path to info file of the chunk. Yields: Dictionaries with images and the different labels. rr rrz{}{:05d}rN) _load_chunk enumeraterzipr&rE) r-r=r>r?r@dat_arrcat_arrinfo_arrircategoryinfo_vecrecordr/r/r0_generate_exampless"   zSmallnorb._generate_examplesN) __name__ __module__ __qualname____doc__r#r(VersionVERSIONSUPPORTED_VERSIONSr1rPr]r/r/r/r0r6s " rcCsXt|}t|d}t|}t|}t|}|dddfd|dddf<|||fS)zLoads a data chunk as specified by the paths. Args: dat_path: Path to dat file of the chunk. cat_path: Path to cat file of the chunk. info_path: Path to info file of the chunk. Returns: Tuple with the dat, cat, info_arrays. NrQ)read_binary_matrixnp expand_dimscopy)r>r?r@ dat_array cat_array info_arrayr/r/r0rSs    rSc Cstjj|d^}|}tdd}ttj ||dd}|dkr)td}n|dkr6tdd}nt d ttj ||dd d }tj |||d d }t d |d }tj ||d |d} Wdn1siwY| t |S)aReads and returns binary formatted matrix stored in filename. The file format is described on the data set page: https://cs.nyu.edu/~ylclab/data/norb-v1.0-small/ Args: filename: String with path to the file. Returns: Numpy array contained in the file. rbint32s"      g