huashu-bird-sticker README

README проекта с github.com/alchaincyf/huashu-bird-sticker. Одним файлом markdown: README.md.

Bird stickers for the world's living bird species

huashu-bird-sticker: specimen drawer, body-size lineup and family plates

花叔开源的全球鸟类贴纸库:按全球统一鸟类名录AviList整理的1.1万多种鸟,每种一张白边贴纸,附多语种鸟名和分布、形态、保护等级数据。贴纸出自花叔做的观鸟App「鸣迹」。中文说明 →

English · 简体中文 · 繁體中文 · 日本語 · 한국어 · Español · Français · Deutsch · Português · Русский · العربية

Bird Album: every bird in the world, one sticker each. birdalbum.com

See the whole library in use at birdalbum.com. Bird Album is a free bird guide built on these stickers: look up any species by name in 15 languages, see where it lives and which months it is around, and open a map of 8,171 birding hotspots that shows what people are seeing at each one this month.

By Huashu (花叔), from the bird-watching app Mingji (鸣迹). More than 11,000 bird species across 45 orders and 249 families, following AviList v2026, the unified global checklist first published in 2025 (the IOC list has ended in its favour). Every species that AviList does not flag as extinct or possibly extinct is included, plus 43 flagged species that already had a sticker: extinct, extinct in the wild, possibly extinct, or, for two, once extinct in the wild and since reintroduced (see extinctionStatus). Every species has a sticker. Each species comes with structured data: taxonomy, names in up to 43 languages, cross-database IDs, and what each sticker was made from.

Version 2.0 moved the taxonomy from IOC 15.2 to AviList v2026. 82 species changed their scientific name (and so their id and sticker path), 91 former species are now subspecies of another species, and 30 new species were split off. Every change is listed in data/crosswalk-ioc15.2-avilist2026.csv, and each record keeps its old IOC ID under previous. See CHANGELOG.

40 stickers from 15 bird orders

AI-generated, and every file says so. Each sticker file carries XMP/IPTC metadata stating that it is an AI-generated illustration, not a photograph (IPTC Digital Source Type trainedAlgorithmicMedia), which AI system was used, that it was not verified by experts and may contain errors, the license, and a link back to this repository.

License in short: the stickers are licensed under CC BY 4.0: use them for anything, including commercial work, editing and redistribution, as long as you credit huashu-bird-sticker with a link to this repository. The data is MIT, and third-party parts are listed in NOTICE.

Use a sticker in one line

Every sticker is served by the jsDelivr CDN (CORS enabled), so you can put it straight into a page:

<img src="https://cdn.jsdelivr.net/gh/alchaincyf/huashu-bird-sticker@main/stickers/Coraciiformes/Alcedinidae/Alcedo_atthis.webp" alt="Common Kingfisher" width="240">

To look a bird up by name, load data/stickers.json (2.7 MB: scientific, English and Chinese name, file path and the species page on birdalbum.com for each species; the 91 subspecies stickers are listed under forms in data/species.json). Names in 43 languages are in data/names.csv.

const idx = await (await fetch("https://cdn.jsdelivr.net/gh/alchaincyf/huashu-bird-sticker@main/data/stickers.json")).json();
const s = idx.stickers.find(x => x.en === "Eurasian Magpie");
const url = idx.cdn + s.file;  // the sticker URL

Credit it next to the image like this, with the link to the species page:

Illustration: [Bird Album](https://birdalbum.com/birds/common-kingfisher/) / [huashu-bird-sticker](https://github.com/alchaincyf/huashu-bird-sticker), CC BY 4.0

Bird of the day

A new bird on your GitHub profile, blog or Notion page every day. Paste this line into any Markdown file (?lang= for 15 languages, ?theme=dark or auto):

[![Bird of the day](https://birdalbum.com/today.svg)](https://birdalbum.com/today/)

Bird of the day

Eighteen of the most colourful and distinctive groups, 24 stickers each, picked so that no two look alike. Every sticker shown here meets the quality bar.

Hummingbirds · Trochilidae — 361 species

Hummingbirds

Parrots & cockatoos · Psittaciformes — 391 species

Parrots & cockatoos

Kingfishers · Alcedinidae — 123 species

Kingfishers

Bee-eaters, rollers & motmots · Meropidae · Coraciidae · Brachypteraciidae · Momotidae · Todidae — 69 species

Bee-eaters, rollers & motmots

Toucans & barbets · Ramphastidae · Capitonidae · Semnornithidae · Megalaimidae · Lybiidae — 132 species

Toucans & barbets

Birds-of-paradise & bowerbirds · Paradisaeidae · Ptilonorhynchidae — 67 species

Birds-of-paradise & bowerbirds

Pittas · Pittidae — 46 species

Pittas

Tanagers · Thraupidae — 392 species

Tanagers

Sunbirds & flowerpeckers · Nectariniidae · Dicaeidae — 207 species

Sunbirds & flowerpeckers

Cotingas & manakins · Cotingidae · Pipridae — 120 species

Cotingas & manakins

Pheasants & partridges · Phasianidae — 185 species

Pheasants & partridges

Ducks, geese & swans · Anatidae — 164 species

Ducks, geese & swans

Pigeons & doves · Columbidae — 338 species

Pigeons & doves

Waxbills & munias · Estrildidae — 138 species

Waxbills & munias

Jays & magpies · Corvidae — 135 species

Jays & magpies

New World orioles & oropendolas · Icteridae — 107 species

New World orioles & oropendolas

Hornbills · Bucerotidae — 64 species

Hornbills

Starlings & mynas · Sturnidae — 118 species

Starlings & mynas

One from each order

One from each order

What's inside

stickers/<Order>/<Family>/<Genus_species>.webp One transparent sticker per species with a white outline, about 1,000 px on the long side. <Genus_species>.<subspecies>.webp are 91 extra stickers of subspecies that used to be separate species
data/species.json One record per species, the complete dataset: taxonomy, names, IDs, range and subspecies, body measurements and ecology, conservation status, sticker and its source. Field reference: data/README.md
data/crosswalk-ioc15.2-avilist2026.csv How every IOC 15.2 species maps to AviList v2026: same, renamed, lumped, split
data/species.csv The same, flattened for spreadsheets
data/names.csv Names in 43 languages (46 columns: Chinese, French and Portuguese each have two variants)
data/languages.json Language codes and the IOC multilingual column each one comes from
data/field-guide.zh-Hans.json Chinese field-guide notes (look, field marks, sexes and ages, similar species) for 1,596 species, CC BY-SA 4.0

A record looks like this (names shortened):

{
  "id": "Accipiter_nisus",
  "scientificName": "Accipiter nisus",
  "authority": "(Linnaeus, 1758)",
  "taxonomy": {"system": "AviList v2026", "seq": 3097, "order": "Accipitriformes", "family": "Accipitridae", "genus": "Accipiter"},
  "names": {"en": "Eurasian Sparrowhawk", "zh-Hans": "雀鹰", "ja": "ハイタカ", "es": "Gavilán común", "fr": "Épervier d'Europe", "…": "…"},
  "ids": {"ebird": "eurspa1", "avibase": "avibase-DB376A66", "wikidata": "Q25380", "gbif": "2480637", "inaturalist": 5106},
  "range": {"text": "lowland to treeline montane forest, open woodland, and urban and suburban parks, from Europe eastward to …", "…": "…"},
  "subspecies": [{"name": "nisus", "authority": "(Linnaeus, 1758)", "range": "breeds Europe to Türkiye and Siberia; winters to Africa"}, "…"],
  "traits": {"massG": 220.8, "wingMm": 207.4, "tailMm": 155.5, "habitat": "Forest", "migration": "partially migratory", "trophicNiche": "Vertivore", "…": "…"},
  "conservation": {"iucn": "LC", "cites": "II", "chinaProtection": "Class II"},
  "taxonConcept": "same",
  "sticker": {
    "file": "stickers/Accipitriformes/Accipitridae/Accipiter_nisus.webp",
    "kind": "ai-illustration",
    "reviewScore": 91,
    "meetsQualityBar": true,
    "license": "CC-BY-4.0",
    "reference": {"type": "photo", "page": "https://commons.wikimedia.org/wiki/File:…", "author": "Dion Art", "license": "CC BY-SA 4.0"}
  }
}

Coverage

How the stickers were made, and how good they are

All 11,102 sticker files are AI-generated illustrations, and every file says so in its metadata (IPTC Digital Source Type trainedAlgorithmicMedia). For each species we wrote down its field marks and found reference photos (since November 2026, where we could find them, a board of photos by several photographers), generated the sticker with GPT-image, upscaled it, and cut it out. A step-by-step explanation with a flowchart and FAQ, in 15 languages: How the stickers are made. If you want to make something similar, see the prompts, pose rules and review rubric in docs/how-the-stickers-are-made.md.

Every sticker was then scored by a separate AI reviewer. In November 2026 we re-checked all of them to a stricter standard, as plates for a field guide: a sticker passes only if identification, plumage, bill and eye, legs and feet, and proportions each score at least 3 out of 4, identification scores 4, and the reviewer would accept it as a field-guide plate. 77 stickers passed as they were; the other 11,025 were generated again until a new version passed.

These are illustrations, not field-guide plates. Treat them as a friendly picture of the bird, and check a field guide before relying on a fine detail. If you spot a mistake, please open an issue — corrections from birders are the most valuable contribution this project can get.

Know a group of birds well? Review one family: tell us which of its stickers are wrong. Each one you flag is regenerated and reviewed again, and you are credited in CONTRIBUTORS.md.

About the reference photos

What a reference is for. A sticker shows what a species looks like: its plumage colours and markings, bill, eye, legs and feet, and proportions. These are facts about the bird, the same things a field guide describes in words. A written description alone is often not enough for the image model to get them right, so we use reference photos.

How we use it. The image model is given a written description of the species' field marks together with reference photos, and generates a new illustration. How the photos were used depends on when a sticker was made. Early stickers, generated before November 2026, used one photo per species, and both the bird's appearance and its pose in that photo served as reference. Since November 2026, the photos are used only to check what the bird looks like, and every bird is drawn in a standard pose that we set, on a plain background (standing side-on for most groups; swimming, flying and a few other groups have poses of their own); where we could find them, the photos come from several photographers (6,369 stickers were made with a board of several images). All but 77 of the early stickers have since been generated again this way; those 77 passed the stricter review as they were. A separate reviewer then checks the result against the field marks. No reference photo is included in this repository.

Credit and removal. The reference photos we recorded for each sticker (page, and the author and license where known) are listed in sticker.reference in data/species.json; stickers generated from a written description alone have "reference": null. A breakdown by reference license is in data/README.md. If one of your photos was used as a reference and you would rather it weren't, open a takedown request and we will replace the sticker within 7 days.

Use it

Raw files can be loaded directly:

https://raw.githubusercontent.com/alchaincyf/huashu-bird-sticker/main/stickers/Accipitriformes/Accipitridae/Accipiter_nisus.webp
import json
species = json.load(open("data/species.json"))["species"]
by_name = {s["scientificName"]: s for s in species}
print(by_name["Pica pica"]["names"]["ja"], by_name["Pica pica"]["sticker"]["file"])

The repository is about 700 MB. To fetch only the data and a few stickers:

git clone --filter=blob:none --sparse https://github.com/alchaincyf/huashu-bird-sticker
cd huashu-bird-sticker && git sparse-checkout set data stickers/Strigiformes

Used by

Using it somewhere? Open an issue and we will add you here.

Credits

If one of your photos was used as a reference for a sticker and you would rather it weren't, open a takedown request and we will replace the sticker within 7 days.

huashu-bird-sticker (alchaincyf/huashu-bird-sticker)

Ищем в базе…

Что это такое?Открытый набор данных со стикерами, нарисованными ИИ, для каждого из 11 064 живых видов птиц. Каждый стикер сопровождается структурированными таксономическими данными и названиями птиц на до 43 языках. Это библиотека стикеров для приложения Mingji для наблюдения за птицами.
Это приложение или только файлы со стикерами и данными?Это просто репозиторий с файлами: изображения стикеров с прозрачным фоном и белым контуром, а также данные в JSON/CSV, но не приложение. Настоящее приложение для наблюдения за птицами, которое использует эти стикеры, — отдельный продукт для iOS под названием Mingji.
Как мне на самом деле пользоваться стикерами и данными?Можно загрузить исходные webp-стикеры и файлы species.json/species.csv прямо из репозитория или клонировать его целиком. Так как весь репозиторий занимает около 700 МБ, если вам не нужны все изображения, можно скачать только папку data и несколько стикеров.
Нужны ли мне какие-то инструменты для работы с файлами?Никакие специальные инструменты не нужны: это просто статичные webp-изображения и данные в JSON/CSV, так что подойдёт любая программа для просмотра изображений или любой парсер JSON/CSV.
Насколько точны стикеры по сравнению с настоящим полевым определителем?Это иллюстрации, созданные ИИ и проверенные по определительным признакам: клюву, глазу, ногам и оперению. Они не так точны, как таблицы в полевом определителе. 7 668 видов соответствуют требованиям проекта к качеству, а остальные уступают в разной степени.
Можно ли использовать эти стикеры в коммерческих целях?Да — стикеры можно использовать для любых целей, в том числе в коммерческих проектах, редактировать и распространять с указанием авторства, кроме создания приложения, сайта или сервиса для наблюдения за птицами, определения видов или полевых определителей. Лицензия на стикеры — PolyForm Shield 1.0.0, а сам набор данных распространяется по лицензии MIT.
Как на самом деле создавались стикеры?Для каждого вида за основу брали эталонное фото (а для 453 сложных видов — набор фотографий с описанием). По нему с помощью GPT-image нарисовали стикер, затем увеличили его и вырезали. Далее отдельный проверяющий оценивал стикер по определительным признакам: клюву, глазу, ногам и оперению.
На каких языках доступны названия птиц?Названия есть на до 43 языках в 46 столбцах. Английский заполнен на 100 %, французский и упрощённый китайский почти полностью, но такие языки, как корейский, арабский, тайский и персидский, пока покрывают лишь от нескольких сотен до примерно тысячи видов. Это ограничено тем, что есть в самом списке IOC.
Что такое Mingji?Mingji (鸣迹) — это приложение-блокнот для наблюдения за птицами на iOS, созданное тем же автором, Huashu. Этот репозиторий — открытая библиотека стикеров и данных, изначально созданная для этого приложения.
Автор
alchaincyf
Размер
1.4 GB
Звёзды
785
Возраст
19 д
Коммиты · 90 дней