필사 모드: Plant ID, Citizen Science & Nature Apps 2026 — iNaturalist / PlantNet / Merlin Bird ID / eBird / BirdNET / PictureThis / Moyamo / Biome Deep Dive

Prologue — In 2026, every roadside wildflower became a data point

May 2026, walking along Yangjaecheon stream in Seoul. Spring flowers are in full bloom. You see a yellow flower. In the old days you would have said "pretty" and walked on. The 2026 smartphone is different. You open the app and snap a photo. Two seconds later the screen answers — "Veronica persica (Persian speedwell). Confidence 96 percent. Common across Korea. Most recent observation: 2 hours ago, Busan." And one more line — "Upload this observation to iNaturalist? It is the global citizen-science database operated by the California Academy of Sciences and National Geographic Society."

This is what nature ID apps are in 2026. Not just a field guide but the **terminal sensor of a global biodiversity monitoring network**. iNaturalist alone holds more than 200 million cumulative observations. Roughly half are "research grade" — meaning two or more identifiers agreed on the species. Academic papers are published on average 12 per week using this data. The casual stroller has, without realizing it, become a supplier to academic data infrastructure.

But the market is split in two. One side is the world of citizen science, non-profit, and open data — iNaturalist, PlantNet, eBird, Merlin Bird ID, BirdNET. The other is SaaS, paid subscription, commercial — PictureThis, PlantSnap, NatureID, Picture Insect. They do the same thing but their philosophy and data-rights model are completely different. And there is one more layer on top — astronomy apps (Sky Guide, Stellarium Mobile), dangerous mushroom apps (ShroomID), and local apps in Korea and Japan (Moyamo, Biome).

Across 18 chapters this post dissects that landscape precisely. What model each app runs on, how accurate it is, how it is priced, and above all — what value it has as citizen-science data. Who should pick what, from kids to researchers, all the way through.

> Models keep converging, and **the operating party makes the difference**. A free app made by Cornell Lab and INRIA and a paid app at three dollars a month from a California startup show you the same result, but where that result flows is different.

1. The 2026 Nature ID App Map — Plants, Birds, Insects, Sky in 4 Categories

Nature ID apps fall roughly into four categories. Each category has a different market leader and data philosophy.

**Category 1 · Plants**

The fiercest competition. iNaturalist (covers all taxa), PlantNet (INRIA, academic), PictureThis (Glority, commercial), PlantSnap, NatureID, Flora Incognita (TU Ilmenau, Germany, academic), and Koreas Moyamo. Accuracy clusters in the 85 to 95 percent range. The difference is cost structure and data-sharing policy.

**Category 2 · Birds**

Surprisingly, one institution dominates — Cornell Lab of Ornithology. Cornell Lab runs eBird (the observation database), Merlin Bird ID (visual, sound, and photo identification), and BirdNET (audio only, jointly with TU Chemnitz). Together they are effectively the global standard. The competitor is Audubon Bird Guide, but it leans more on field-guide depth and US regional data than on identification.

**Category 3 · Insects**

The hardest. The number of species is overwhelming (estimated over one million), and morphological variation is huge. iNaturalist’s Seek mode is most familiar to general users but weak at fine identification. Commercially, Picture Insect (Glority’s insect title) exists. Korea has Dapago, Japan has Biome — both cover insects.

**Category 4 · Astronomy**

A different character from the other categories. Not identification but AR apps that show "what star, planet, or satellite is in this direction right now." Sky Guide (iOS, Fifth Star Labs), SkyView (Terminal Eleven), Stellarium Mobile (mobile of the open-source desktop Stellarium), Star Walk 2 (Vito Technology), and Star Chart (Escapist Games) are the leaders.

There are also **mushrooms** and **mammals**. Mushrooms are dealt with separately because of edibility/toxicity risk. Mammals are hard to observe directly, so camera-trap projects (Mammal Web, Snapshot Serengeti) are the main method.

This post walks them top to bottom. The key — **free is not all the same free, and two apps doing the same thing can differ five to tenfold in accuracy and data reliability.**

2. iNaturalist — The Citizen Science Standard (200M+ Observations)

iNaturalist began in 2008 as a UC Berkeley master’s project. In 2014 it joined the California Academy of Sciences, and in 2017 the National Geographic Society became a co-sponsor. From 2024 it became an independent non-profit, while keeping the same backers. As of May 2026, cumulative observations passed 210 million. Nearly half are classified as "research grade," meaning two or more identifiers agreed on the same species.

**How it runs**

A user uploads a photo. A computer-vision model proposes three to five candidates. The model is a self-trained CNN (the migration from a ResNet baseline to a Vision Transformer one was completed in 2025). But it does not stop there. The observation is **shared with the community**, and other users (including experts) can add or correct identifications. When two or more agree on a species, it becomes "research grade." When the model is wrong, humans fix it. That data then feeds back into model training.

**Data rights**

The user chooses the license. The default is CC-BY-NC (attribution, non-commercial). It can be opened further to CC0 or CC-BY. The data syncs weekly to GBIF (Global Biodiversity Information Facility), from which it flows to researchers, governments, and NGOs worldwide. In other words **a single photo you upload becomes a unit datum in academic databases**.

**Language and region**

Strong multilingual support. Korean, Japanese, Chinese, Spanish, Portuguese — over 35 languages. In Korea it has a partial cooperative relationship with the Ministry of Environment (not yet a formal integration). In Japan there are regional guides built with the University of Tokyo museum.

**Weaknesses**

Three. (1) Fine identification still needs human hands. The model is an assist. (2) The data is biased — toward cities, developed countries, and sunny weekends. There is no Siberian winter data. (3) The UI is heavy for kids and beginners. That is why Seek exists separately.

**Who should use it**

Serious nature enthusiasts, students, anyone who wants to contribute to citizen science. If you intend to use data academically, this is the first choice.

3. Seek — iNaturalist’s Gamified Version for Kids

Seek is a sibling app iNaturalist launched in 2018. Three core differences.

**Difference 1 · No account required**

You start without signing up. This is because of children’s online privacy protection (COPPA, the US law). Observations stay on the device. You can upload to an iNaturalist account if you want, but local is the default.

**Difference 2 · Gamification**

A badge system — "observe ten mammals," "observe fifty flowers." Monthly challenges ("April flower challenge"). Designed for kids to take nature observation as a game. From a parent or teacher’s perspective, it is a strong learning tool.

**Difference 3 · Same model, different confidence threshold**

It uses the same CNN/ViT model but Seek only answers "when sure." When uncertain it gives only the higher taxon (e.g., "Lamiaceae"). The design prevents children from memorizing incorrect identifications.

**Weaknesses**

The badges and challenges are shallow. Observing the same species multiple times does not count as new data. As a game it is thin. Accuracy is also somewhat more conservative than mainline iNaturalist (because of the higher threshold).

**Who should use it**

Families with kids aged 6 to 14, elementary-school nature classes, nature-camp leaders. If you are an adult, mainline iNaturalist is the better choice.

4. PlantNet (INRIA France) — A Model Academic Project

PlantNet is co-operated by France’s national informatics and applied math institute (INRIA) together with CIRAD (international agricultural research center), IRD (development research), and Tela Botanica (citizen-botanist network). Released in 2013. As of 2026, more than 15 million cumulative users and 100 million observations. The key — because an academic institution built it, **researchers can know how the model was trained**. The dataset, model weights, and evaluation protocol are all open.

**Technical structure**

A CNN-family model fine-tuned on PlantNet’s own dataset (about 6 million labeled images) on top of an ImageNet-pretrained backbone. Migration to a ViT-B/16 base started in 2024. The 2026 model identifies more than 45,000 plant species worldwide. Accuracy is over 95 percent for regional floras like Western Europe, but drops into the 70s for parts of tropical Asia and Africa. This is dataset bias (European data overwhelmingly dominates).

**Pl@ntNet vs PlantNet**

Same app. The official spelling is Pl@ntNet (with the @). For pronunciation and search convenience, PlantNet is also commonly used. This post sticks with the common form PlantNet.

**Project system**

PlantNet’s differentiator. You can create a project for a specific region or theme (e.g., "Mediterranean invasive species," "Madagascar endemics") and classify and collect data inside it. Useful for academic surveys, environmental monitoring, and citizen-science campaigns.

**Data sharing**

All observations are by default available for academic use. Syncs to GBIF. Unlike iNaturalist there are fewer license options (an academic-use consent default).

**Weaknesses**

(1) The community is not as active as iNaturalist — fewer ID discussions. (2) Birds and insects are not covered — strictly plants. (3) The UI is somewhat academic — cold for general users.

**Who should use it**

University students, botany majors, agriculture workers, anyone who needs a tool with academically validated accuracy. Pair it with iNaturalist for cross-validation.

5. PictureThis (Glority) — Leader of the Commercial Market

PictureThis is operated by Glority, based in California. Released in 2017, it has been the fastest-growing commercial plant ID app. As of 2026, more than 200 million cumulative downloads and 50 million monthly active users. Almost always rank one in the plant category on the App Store. It is not free — a seven-day free trial then a $3/month or $30/year subscription is the default. The free tier has daily ID limits and ads.

**Why so popular**

Three reasons. (1) **The UX is overwhelming.** Snap a photo, done. ID, detailed info (origin, toxicity, growing tips), similar species, and care guide — all in one screen. iNaturalist is academic; PictureThis is gardener-friendly. (2) **Multilingual is genuinely good.** Korean, Japanese, Chinese interfaces feel nearly equivalent to English. (3) **Fast and accurate — for common garden plants.** 95 percent accuracy in garden, indoor, and urban-common species.

**Limits**

For wildflowers, rare species, and regional natives, accuracy drops. The academic range covered by PlantNet and iNaturalist is not covered by PictureThis. **Data is private** — user observations do not flow to academic databases. They are Glority’s proprietary asset.

**Glority’s other apps**

The same company runs Picture Insect (insects), Picture Bird (birds), Picture Mushroom (mushrooms), Rock Identifier (rocks), Healthy Cat/Dog (pet health). PictureThis is essentially the plant title of the "Picture series." Many share the same backend and subscription.

**Pricing details**

PictureThis Premium is $3/month or $30/year. A lifetime subscription ($80) exists. There is a separate Family Plan (shared with family). Auto-renewal after a seven-day free trial — many App Store reviews complain about people forgetting and being charged, leading to refund disputes.

**Who should use it**

Gardeners, indoor-plant keepers, general users who need plant-care info. If you also need academic accuracy, pair with PlantNet/iNaturalist.

6. PlantSnap / NatureID / Flora Incognita — Other Plant Apps

Beyond the first-tier commercial and academic apps, several second-tier apps occupy slightly different positions.

**PlantSnap (Earth.com)**

Released 2017. Under US-based Earth.com. Marketed aggressively ("600,000 species identified") and reached 40 million downloads, but lost ground to PictureThis from the mid-2020s. UI feels somewhat dated, the free/paid split is confusing. User-comparison reviews put it 5 to 10 percent below PictureThis on average accuracy.

**NatureID**

Run by Next Vision Limited in the UK. Plants + animals + birds + insects + mushrooms in one app. Aimed at "all-in-one" but falls short of category leaders. The UI is clean and supports Korean/Japanese. Pricing is similar to PictureThis ($3.99/month).

**Flora Incognita (TU Ilmenau, Germany)**

An academic project from Ilmenau University of Technology and the Max Planck Institute for Chemical Ecology. Released 2018. Free. Similar academic positioning to PlantNet but specialized in German and Central European flora. About 4,800 species covered. After identification it also shows IUCN Red List status and protected-species flags. Often recommended for environmental education.

**LeafSnap**

Released in 2011. Built jointly by Columbia, Maryland, and Smithsonian — an early model based on leaf shape. By the 2020s it is essentially in maintenance. Historically important but with few active users in 2026.

**Selection guide**

For general users, PictureThis (paid) or PlantNet (free). For academic, PlantNet + iNaturalist. If you live in Germany or Central Europe, add Flora Incognita. If you want one app for all taxa, NatureID is worth trying.

7. eBird (Cornell Lab) — The Global Standard for Bird Observation

eBird is a citizen-science project launched in 2002 by the Cornell Lab of Ornithology together with the Audubon Society. As of 2026, cumulative observation data exceeds 1.7 billion records. Two to three hundred million records are added each year. It is the largest source of data for estimating global bird distribution, migration, and population. More than twenty academic papers a week are published using it.

**How it runs**

A user (usually a serious birder) uploads a **checklist** after a walk or birding session. The checklist records "where, when, how long, what, how many." It is not simply photo ID — **counting individuals is the core**. A veteran birder reporting that they saw 30 species totaling 50 individuals over one hour quantifies the bird activity at that time and place.

**Hotspot system**

Famous birding spots worldwide are registered as "Hotspots." Where to go to see what bird, the seasonal probability — all built from citizen data.

**eBird Status & Trends**

Academic publication built by Cornell Lab on top of eBird data. Per species, annual distribution, migration, and population change visualizations. Free and public. The basis of conservation policy.

**Weaknesses**

(1) **No auto-identification.** eBird is a system where the user types the species in. For auto-ID, you need Merlin Bird ID separately (see chapter 8). (2) **Serious birder centric.** General users face a learning curve. (3) The UI is heavy because it is data-entry centric.

**Who should use it**

Serious birders, ornithology majors, conservation workers, and anyone who wants to contribute their observations to global academic data. Korea is growing as well (about 8,000 active users in 2026).

8. Merlin Bird ID — Visual + Sound + Photo Integrated

Merlin Bird ID is also a Cornell Lab product. If eBird specializes in "observation records," Merlin specializes in "identification." Released 2014. As of 2026, more than 50 million cumulative downloads. Free. Ad-free. Within Cornell Lab’s non-profit mission.

**Three ID modes**

(1) **Photo ID** — single-photo ID, similar to iNaturalist. (2) **Sound ID** — show it 5 to 10 seconds of birdsong, get the species. The BirdNET technology (chapter 9) is integrated here. Since the 2021 launch this has been the most popular feature. (3) **Step-by-Step ID** — the user answers size, color, and behavior questions step by step to narrow possibilities. The old field-guide method.

**Sound ID is the game changer**

Birds are hard to see but easy to hear. Sound ID listens in the background when you leave the mic on and shows incoming birds in real time. Watching the dawn chorus light up with seven or eight species at once on a single screen — astonishing. Accuracy is in the 90s for common species and 60 to 70 percent for rarities.

**Bird Pack system**

Regional models are downloaded as separate packs. "South Korea Pack," "Japan Pack" and so on. Offline use after download. A pack is usually 200 to 500 MB.

**eBird linkage**

You can add a bird identified in Merlin to an eBird checklist in one tap. Identification, recording, global database — one flow.

**Weaknesses**

(1) **Sound ID struggles with environmental noise.** Cities, wind, traffic noise drop accuracy. (2) **Rare and look-alike species still need human verification.** The small flycatchers in genus Empidonax are routinely mis-IDed. (3) Some regional packs lack data.

**Who should use it**

From beginning birders to veterans. Bird-ID accuracy and UX that surpass iNaturalist. **If you are curious about birds, download this first.**

9. BirdNET (Cornell + TU Chemnitz) — Audio-only Identification

BirdNET is an audio-based birdsong ID model released in 2019 by Cornell Lab together with the Stefan Kahl research team at TU Chemnitz in Germany. First released as an academic model, the consumer BirdNET mobile app (free) followed in 2021. At the same time, BirdNET-Analyzer was released on GitHub as an open-source tool that researchers apply to their own recordings.

**Model architecture**

A CNN trained on a birdsong dataset (mostly Xeno-canto and the Macaulay Library) on top of an EfficientNet backbone. Audio is converted to mel spectrograms and treated as an image classification problem. A three-second sliding window. Per window, the model outputs candidate species with confidence scores.

**Mobile app**

The UX is similar to Merlin Sound ID. In fact Merlin Sound ID was built using BirdNET. The difference — the BirdNET app is **researcher and engineer friendly**. You can adjust the confidence threshold directly, and export results as raw CSV. Merlin is user-friendly; BirdNET is data-friendly.

**BirdNET-Pi**

A community project to run BirdNET on a Raspberry Pi for 24-hour birdsong monitoring is popular. Installed in gardens or research sites for automated data collection. Over 20,000 GitHub stars.

**Coverage**

Over 6,000 species worldwide. Strong in North America and Europe; East Asia improves incrementally. About 300 Korean bird species are in the model.

**Weaknesses**

(1) **No visual ID.** Audio only. Pair with Merlin or iNaturalist. (2) **Weak at low-frequency and wind noise.** Difficult in urban environments. (3) The UI leans somewhat academic — not entirely friendly for beginners.

**Who should use it**

People interested in avian bioacoustics, those who want to set up birdsong monitoring in a garden, anyone who needs to export data academically. General birders are well served by Merlin Sound ID.

10. Audubon Bird Guide — The American Classic

Audubon Bird Guide is the app of the National Audubon Society, the oldest bird-conservation group in the US (founded 1905). The app launched in 2010. More than 10 million cumulative downloads as of 2026. Free.

**Positioning**

Compared to Merlin, Audubon is **closer to a field guide than to an ID tool**. The most complete digital field guide covering 821 species in northern North America. For each species — illustrations, photos, range maps, four-season patterns, behavior, similar-species comparisons, audio samples (including variants).

**Auto-ID features**

Photo ID and Sound ID exist but are generally less accurate than Merlin/BirdNET. If ID is the core need use Merlin; pairing Audubon as a field guide is the US birder standard.

**Audubon Migration Explorer**

An interactive visualization built on Audubon’s own data plus eBird. Shows annual migration routes per species. A powerful conservation-education and policy tool.

**Weaknesses**

US-Canada-Mexico centric. Asia, Europe, South America are essentially uncovered. In Korea, only partially useful (some overlap with winter migrants).

**Who should use it**

Birders living in the US and Canada, anyone who wants a US bird field guide digitally, people who want to combine English learning with nature learning. For other regions, low priority.

11. Sky Guide / SkyView / Stellarium Mobile / Star Walk 2 — Astronomy

Astronomy apps are the most mature segment of the ID category. AR (augmented reality) tech worked well from the early 2010s. Point the phone at the sky and stars, planets, satellites, galaxies, and artificial satellites are labeled in real time.

**Sky Guide (Fifth Star Labs, iOS)**

Released 2014. iOS only. $2.99 one-time purchase (no subscription). Widely praised for the most beautiful design. Constellation illustrations, dark mode, satellite tracking (ISS, Starlink), celestial event alerts. User average rating 4.9. **Effectively the default for iOS users.**

**SkyView (Terminal Eleven)**

iOS and Android. Free tier plus Pro at $1.99. Praised for the most stable AR. Strong at tracking satellites — ISS, Hubble, Starlink in real time. One-time payment, no subscription fatigue.

**Stellarium Mobile (Noctua Software)**

Mobile port of desktop Stellarium. Free tier plus Plus at $9.99/year. Desktop Stellarium is the standard open-source planetarium software. The mobile version is **the deepest tool for astronomy majors and amateur astronomers**. Catalog is overwhelming — 19 million stars, 1 million deep-sky objects. Telescope control integration too.

**Star Walk 2 (Vito Technology)**

iOS and Android. Free tier plus Pro at $2.99. The most family-friendly UI. Frequently recommended for kids and families. Strong "what to look at tonight" guides on top of AR. Time travel (past/future sky simulation) too.

**Star Chart (Escapist Games)**

iOS and Android. Free plus in-app purchases. One of the oldest sky AR apps (2010). Still popular. The UI feels dated but is stable.

**Summary comparison**

- **iOS, design first** → Sky Guide

- **Android, AR stability** → SkyView

- **Astronomy depth/majors** → Stellarium Mobile

- **Family/kids, guides** → Star Walk 2

- **Free/ads OK** → Star Chart

This category does not collect citizen-science data — measurements come from the observatory side. The user is "watcher," not "recorder."

12. Insect Identification — Picture Insect / Seek / iNaturalist

Insects are the hardest taxon. Too many species (more than one million estimated, with about 900,000 formally classified), large morphological variation, hard to photograph. Even so, ID apps keep improving.

**iNaturalist + Seek**

Most familiar pair for general users. iNaturalist’s strength is the expert community helping with IDs. Entomologists are active. Seek is good for kids with gamification.

**Picture Insect (Glority)**

The insect title of PictureThis. $3/month or $30/year. More than 4,000 species. Clean fast UI. Over 90 percent for common insects (ants, ladybugs, common butterflies). But for tough orders like small flies, bees, and moths, accuracy drops. Pair with iNaturalist for safety.

**Insect Identifier**

Several small apps that emerged from 2020 onward. Variable accuracy. Caution recommended.

**Korea and Japan**

Korea has Dapago; Japan has Biome (chapters 14, 15). Both cover insects too.

**Hard subfields — spiders and myriapods**

Spiders often cannot be IDed to species by any app. The best route — upload to iNaturalist and wait for an arachnologist to ID it.

**Who should use it**

- Kids/beginners → Seek

- General use with accuracy first → iNaturalist + Picture Insect together

- Living in Korea → Dapago

- Living in Japan → Biome

- Specialists → iNaturalist + taxon-specific expert communities

13. Mushroom Identification — Risk + ShroomID / PictureMushroom

A warning first. **Do not eat wild mushrooms based on app output.** Mushrooms with nearly identical Latin names and external appearance can be edible or deadly. The death cap (Amanita phalloides) and the edible Volvopluteus are hard to tell from a single photo. From the 2010s onward, several people have died after trusting an ID app and eating wild mushrooms. The US CDC and EFSA both officially advise that **app output should not be used as a basis for edibility decisions.**

**Why they exist anyway**

For observation, learning, and recording, they have value. Wanting to know "what is this mushroom" is legitimate curiosity. Eating is a wholly separate matter.

**ShroomID / Picture Mushroom (Glority)**

Glority’s mushroom title. UX identical to PictureThis. Free/paid structure identical ($3/month). Edibility/toxicity flags appear, but the app itself shows a disclaimer that "this information should not be used to decide on edibility."

**Mushroom World**

An older academic-field-guide app. Reputed for somewhat higher accuracy. Identification is weak but useful as a field guide.

**iNaturalist**

The mushroom expert community is active. Uploading a photo often gets a mycologist to ID it. It takes time but accuracy is best.

**Hard genera like Boletes and Russula**

A photo alone cannot ID to species in many genera. DNA sequencing is sometimes required. A safe app design tells you only the genus.

**Who should use it**

- Observation/recording only → iNaturalist

- Field-guide learning → Mushroom World + iNaturalist

- **Never make edibility decisions from an app — go through a local expert or association**

14. Korea — Moyamo / Dapago / Ministry of Environment Nature Guides

The Korean market sees global apps and local apps coexist. Local apps are stronger in identifying Korean natives and in Korean-language information depth.

**Moyamo**

Released 2017. The de facto standard for Korean plant ID and community. The user uploads a photo, the computer vision proposes candidates, and the community (especially expert plant enthusiasts and gardeners) helps confirm the ID. For Korean plants, many user comparisons rate it higher than PictureThis — because the dataset of Korean natives and naturalized species is rich. Free with ads, some premium features. Unlike iNaturalist, the data circulates mainly within the Korean user community (sync with global academic data is partial).

**Dapago**

A broader nature ID app. Covers plants but also insects and birds. Where Moyamo is plant-centric, Dapago is wider taxonomically. It has links to the Korean Ministry of Environment’s natural-environment surveys. The UI is somewhat heavy, but for Korean native insects, it is often more accurate than global apps.

**Ministry of Environment / National Institute of Biological Resources**

The Ministry of Environment and the National Institute of Biological Resources provide official nature observation resources. More of a web-based field guide than an app (species.nibr.go.kr). Information on endangered species, protected wild plants, and natural monuments is the most accurate. The most trustworthy source for verifying Latin and Korean common names.

**iNaturalist Korea**

Korean users on the global iNaturalist platform are growing. About 2 million cumulative observations in 2026. Many users pair it with Moyamo.

**Korean birds — eBird / Merlin**

Korean bird observation is effectively standardized on global eBird/Merlin. Synchronized with the Korean Ornithologists Association’s bird checklist.

**Who should use what**

- Korean plant ID → Moyamo (first), iNaturalist (auxiliary)

- Korean insect ID → Dapago, iNaturalist

- Korean birds → Merlin Bird ID + eBird

- Latin and legally protected species → Ministry of Environment / NIBR species.nibr.go.kr

- Global academic contribution → iNaturalist main

15. Japan — Biome / Picture Insect Japan / Gakken Zukan LIVE

Japan’s nature-ID app market is deeper than Korea’s. The culture of educational manga and field guides is strong, and a population of 100 million makes a stable market.

**Biome (Japan biodiversity)**

Released in 2019. A Japanese citizen-science biodiversity platform. Covers plants, insects, birds, fish, amphibians, reptiles, and fungi — all taxa. Japanese data is overwhelming. Strong gamification — missions, badges, species collection. Very popular with Japanese kids and teens. Linked to the Ministry of the Environment’s nature observation campaigns. Some data is shared with the Ministry of the Environment and the National Institute for Environmental Studies. Some Korean and Chinese users exist, but data outside Japan is sparse.

**Picture Insect Japan**

The Japanese localization of Glority’s Picture Insect. The Japanese insect dataset is added. Popular for Japanese kids’ summer free-research projects. Same structure as the Korean version plus Japanese UI.

**Gakken Zukan LIVE**

Gakken is a major Japanese educational publisher. "Gakken Zukan LIVE" is the digital companion app for the paper field-guide series. AR overlays video on the field-guide pages, and there is some photo-ID functionality. Accuracy is not on par with global apps, but as an educational tool for kids it is overwhelming. The book + app + video integrated experience is specialized for the Japanese family market.

**Yamap**

An app combining hiking and nature observation. The Japanese hiker standard. You can record plants and birds you met along with your hiking log. Meaningful as citizen-science data.

**Other Japanese apps**

Japan’s paper field guide culture is still strong. Using paper guides alongside digital apps is standard in Japan. Region-specialized apps like Kanagawa’s "river and ocean life" guide are also numerous.

**Who should use what**

- Living in Japan, kids education → Gakken Zukan LIVE + Biome

- Living in Japan, general identification → Biome + iNaturalist

- Japanese hikers → Yamap

- Japanese birds → Merlin Bird ID (Japan pack) + eBird

16. The Meaning of Citizen Science — Data Accuracy + Ethics

Now step back. How does our act of uploading a photo on a walk become academic data, and what responsibility comes with it?

**Data accuracy**

There are several academic evaluations showing the iNaturalist "research grade" data has an average accuracy above 95 percent (typically 90 to 98 percent). But the variance by taxon is large. Birds average 97 percent, plants 92 percent, insects 85 percent, fungi below 80 percent. Trustworthiness depends on which taxon you are looking at.

**Bias problem**

The biggest weakness of citizen-science data is **sampling bias**. Data clusters in cities, developed countries, sunny weekends, and popular species. Siberian winter fungi, Saharan insects, rainy-day birds are absent. For academic use, this bias must be statistically corrected. Sophisticated models like eBird Status & Trends handle this.

**Ethics of location data**

Publishing the exact GPS of an endangered species turns it into a target for poaching or theft. If you carelessly publish the location of endangered Korean plants (e.g., kashioh, Halla-songii-pul) or rare insects, someone may go collect them. iNaturalist provides "obscured location" to prevent this — endangered species are automatically blurred onto a roughly 10 km grid. Users can manually blur locations too. **Always obscure locations when dealing with rare species.**

**Ethics of observation**

Do not approach a nest just to take a photo, do not pick a plant, do not chase an animal. The first principle of citizen science is — **do not influence the subject of observation**. If you get close enough that a bird flees, the observation itself becomes a variable.

**Copyright and license**

Be conscious of license when uploading photos. The iNaturalist default CC-BY-NC permits non-commercial use. CC0 waives all rights. You decide the fate of your own photo.

**AI model training data**

Photos you upload become training data for the next model version — if the license allows. This is a positive cycle between citizen science and AI, but some users disagree with their photos being used in training. The license choice determines that.

17. Who Should Pick What — Hobbyists / Students / Research / Family

Finally, recommendations by user type.

**Type 1 · Light hobbyist (weekend walker)**

- Plants → PictureThis (convenient, paid) or PlantNet (free)

- Birds → Merlin Bird ID

- Insects → Seek

- Sky → Sky Guide (iOS) or SkyView (Android)

- Living in Korea → add Moyamo

- Living in Japan → add Biome

**Type 2 · Kids/family (nature learning)**

- Main → Seek (gamification, safety, free)

- Sky → Star Walk 2 (kid-friendly)

- Living in Japan → Gakken Zukan LIVE

- School nature observation → iNaturalist with a parent

**Type 3 · Serious hobbyist (weekend birder/nature club)**

- Main → iNaturalist

- Bird specialist → Merlin Bird ID + eBird

- Plant complement → PlantNet

- Korean plants → Moyamo

**Type 4 · Undergraduate/graduate (academic)**

- Data contribution → iNaturalist + PlantNet

- Birds → eBird + BirdNET-Analyzer

- Audio analysis → BirdNET-Analyzer (CLI)

- Regional data → Ministry of Environment material (Korea), Biome (Japan)

**Type 5 · Specialist/researcher**

- Main data source → GBIF (iNaturalist and eBird integrated)

- Taxon specialist → iNaturalist expert communities

- Bird data → eBird Status & Trends

- Acoustic monitoring → BirdNET-Pi or BirdNET-Analyzer

- Regional native → Moyamo (Korea), Biome (Japan), Flora Incognita (Germany)

**Type 6 · Environmental educator (teacher, camp leader)**

- Main → Seek + iNaturalist combo

- Living in US → add Audubon Bird Guide

- Data-driven activity → eBird or PlantNet projects

- Living in Japan → Gakken Zukan LIVE + Biome

**Type 7 · Gardener/urban naturalist**

- Plant care → PictureThis (rich care info)

- Garden birds → Merlin Sound ID (birds in the yard)

- Garden insects → Picture Insect

- Garden monitoring → BirdNET-Pi (DIY Raspberry Pi)

**Decision tree one-liner**

- Free + academic → iNaturalist

- Paid + convenient → PictureThis

- Bird-only → Merlin Bird ID + eBird

- Kids → Seek

- Korea → Moyamo + iNaturalist

- Japan → Biome + iNaturalist

- Sky → Sky Guide / SkyView / Stellarium

References

- iNaturalist: https://www.inaturalist.org/

- iNaturalist Network (regional nodes): https://www.inaturalist.org/pages/network

- California Academy of Sciences: https://www.calacademy.org/

- Seek by iNaturalist: https://www.inaturalist.org/pages/seek_app

- Pl@ntNet (PlantNet): https://plantnet.org/

- INRIA (Institut national de recherche en sciences et technologies du numérique): https://www.inria.fr/

- PictureThis: https://www.picturethisai.com/

- Glority: https://glority.com/

- PlantSnap: https://www.plantsnap.com/

- NatureID: https://natureid.com/

- Flora Incognita: https://floraincognita.com/

- TU Ilmenau Flora Incognita project: https://www.tu-ilmenau.de/en/florain

- LeafSnap: https://leafsnap.com/

- eBird: https://ebird.org/

- Cornell Lab of Ornithology: https://www.birds.cornell.edu/

- Merlin Bird ID: https://merlin.allaboutbirds.org/

- BirdNET: https://birdnet.cornell.edu/

- BirdNET-Analyzer GitHub: https://github.com/kahst/BirdNET-Analyzer

- BirdNET-Pi GitHub: https://github.com/mcguirepr89/BirdNET-Pi

- TU Chemnitz Media Informatics (BirdNET researchers): https://www.tu-chemnitz.de/informatik/Medieninformatik/

- Audubon Bird Guide: https://www.audubon.org/app

- National Audubon Society: https://www.audubon.org/

- Sky Guide (Fifth Star Labs): https://www.fifthstarlabs.com/sky-guide

- SkyView (Terminal Eleven): https://www.terminaleleven.com/skyview/

- Stellarium Mobile: https://stellarium-labs.com/stellarium-mobile-plus/

- Star Walk 2 (Vito Technology): https://starwalk.space/

- Star Chart (Escapist Games): https://www.escapistgames.com/sc.html

- Picture Insect: https://www.pictureinsect.com/

- Picture Mushroom: https://www.picturemushroom.com/

- Mushroom World: http://www.mushroom.world/

- Moyamo: https://www.moyamo.kr/

- Korea NIBR species search: https://species.nibr.go.kr/

- Korea Ministry of Environment: https://www.me.go.kr/

- Biome (Japan): https://biome.co.jp/app-biome/

- Gakken Zukan LIVE: https://zukan.gakken.jp/live/

- Yamap (Japan): https://yamap.com/

- GBIF (Global Biodiversity Information Facility): https://www.gbif.org/

- Xeno-canto (birdsong dataset): https://xeno-canto.org/

- Macaulay Library (Cornell Lab): https://www.macaulaylibrary.org/

- IUCN Red List: https://www.iucnredlist.org/

- US CDC wild mushroom warning: https://www.cdc.gov/

- EFSA: https://www.efsa.europa.eu/

- Mammal Web (UK camera-trap citizen science): https://www.mammalweb.org/

- Snapshot Serengeti: https://www.snapshotserengeti.org/

- iSpot (UK nature ID): https://www.ispotnature.org/