keremberke/pokemon-classification

该数据集属于图像分类任务，包含各种宝可梦的图像。数据集通过Roboflow平台导出，包含6991张图像，分为训练集（4869张）、验证集（1390张）和测试集（732张）。每张图像经过自动定向和调整大小至224x224像素的处理。数据集标签包括Porygon、Goldeen、Hitmonlee等多种宝可梦名称。

原始信息汇总

数据集概述

数据集标签

• 包含以下标签：[Porygon, Goldeen, Hitmonlee, Hitmonchan, Gloom, Aerodactyl, Mankey, Seadra, Gengar, Venonat, Articuno, Seaking, Dugtrio, Machop, Jynx, Oddish, Dodrio, Dragonair, Weedle, Golduck, Flareon, Krabby, Parasect, Ninetales, Nidoqueen, Kabutops, Drowzee, Caterpie, Jigglypuff, Machamp, Clefairy, Kangaskhan, Dragonite, Weepinbell, Fearow, Bellsprout, Grimer, Nidorina, Staryu, Horsea, Electabuzz, Dratini, Machoke, Magnemite, Squirtle, Gyarados, Pidgeot, Bulbasaur, Nidoking, Golem, Dewgong, Moltres, Zapdos, Poliwrath, Vulpix, Beedrill, Charmander, Abra, Zubat, Golbat, Wigglytuff, Charizard, Slowpoke, Poliwag, Tentacruel, Rhyhorn, Onix, Butterfree, Exeggcute, Sandslash, Pinsir, Rattata, Growlithe, Haunter, Pidgey, Ditto, Farfetchd, Pikachu, Raticate, Wartortle, Vaporeon, Cloyster, Hypno, Arbok, Metapod, Tangela, Kingler, Exeggutor, Kadabra, Seel, Voltorb, Chansey, Venomoth, Ponyta, Vileplume, Koffing, Blastoise, Tentacool, Lickitung, Paras, Clefable, Cubone, Marowak, Nidorino, Jolteon, Muk, Magikarp, Slowbro, Tauros, Kabuto, Spearow, Sandshrew, Eevee, Kakuna, Omastar, Ekans, Geodude, Magmar, Snorlax, Meowth, Pidgeotto, Venusaur, Persian, Rhydon, Starmie, Charmeleon, Lapras, Alakazam, Graveler, Psyduck, Rapidash, Doduo, Magneton, Arcanine, Electrode, Omanyte, Poliwhirl, Mew, Alolan Sandslash, Mewtwo, Weezing, Gastly, Victreebel, Ivysaur, MrMime, Shellder, Scyther, Diglett, Primeape, Raichu]

图像数量

• 训练集：4869张
• 验证集：1390张
• 测试集：732张

使用方法

• 安装 datasets 库： bash pip install datasets

• 加载数据集： python from datasets import load_dataset ds = load_dataset("keremberke/pokemon-classification", name="full") example = ds[train][0]

数据集处理

• 每张图像应用了以下预处理：
  • 自动调整像素数据方向（去除EXIF方向信息）
  • 调整大小至224x224（适应性调整，保留黑色边缘）
• 未应用图像增强技术。