Ollama × Gemma × Pokemon API で画像認識を試す

import { readFile, writeFile } from 'node:fs/promises'

const OLLAMA_URL = 'http://localhost:11434/api/chat'
const MODEL = 'gemma3:4b'
const CONCURRENCY = 2
const DATA_PATH = new URL('../src/data/pokemon-gen1-3.json', import.meta.url)

const PROMPT = `この画像は透明背景のポケモンのドット絵スプライトです。
このポケモンは地面から離れて宙に浮いている・浮遊している・飛行している状態ですか？
それとも足や体が地面についている状態ですか？
次のJSON形式だけで答えてください: {"floating": true または false}`

async function fetchImageBase64(url) {
  const res = await fetch(url)
  const buf = Buffer.from(await res.arrayBuffer())
  return buf.toString('base64')
}

async function labelOne(entry) {
  const imageUrl = entry.officialArtworkUrl || entry.imageUrl
  if (!imageUrl) return null

  const image = await fetchImageBase64(imageUrl)
  const res = await fetch(OLLAMA_URL, {
    method: 'POST',
    headers: { 'Content-Type': 'application/json' },
    body: JSON.stringify({
      model: MODEL,
      stream: false,
      format: { type: 'object', properties: { floating: { type: 'boolean' } }, required: ['floating'] },
      messages: [{ role: 'user', content: PROMPT, images: [image] }],
    }),
  })
  const data = await res.json()
  const parsed = JSON.parse(data.message.content)
  return parsed.floating
}

async function main() {
  const entries = JSON.parse(await readFile(DATA_PATH, 'utf-8'))
  const retryOnly = process.argv.includes('--retry-failed')
  const targets = retryOnly ? entries.filter(e => e.floating == null) : entries
  console.log(`processing ${targets.length}/${entries.length} entries${retryOnly ? ' (retry-failed mode)' : ''}`)

  let cursor = 0
  async function worker() {
    while (cursor < targets.length) {
      const idx = cursor++
      const entry = targets[idx]
      try {
        entry.floating = await labelOne(entry)
      } catch (err) {
        entry.floating = null
        console.error(`\n${entry.name} failed: ${err.message}`)
      }
      process.stdout.write(`\r${idx + 1}/${targets.length} (${entry.name}: ${entry.floating})`.padEnd(60))
    }
  }
  await Promise.all(Array.from({ length: CONCURRENCY }, worker))
  console.log()

  await writeFile(DATA_PATH, JSON.stringify(entries, null, 2))
  console.log(`updated ${entries.length} entries in ${DATA_PATH.pathname}`)
}

main()