🖼️ ビジョン・画像ファミリー

コンピュータビジョン・画像AI API

Name: Brainiall Vision & Image AI
Brand: Brainiall
SKU: vision

EC商品用の微細な髪の毛まで切り抜く背景透過、超解像ディテール復元、顔写真の鮮明化。

無料トライアルを開始

高精度背景切り抜き（アルファマット）

POST/v1/vision/remove-background

髪の毛やガラスの透過エッジを忠実に保持しながら背景を自動除去します。

Pixel-level alpha matting preserving fine hair, fur, and transparent objects
Returns transparent PNG, WebP, or custom RGB background composites
High-throughput processing under 400ms per 4K image

ResolutionUp to 4096 x 4096 px

Output FormatsPNG, WebP (RGBA)

POST /v1/vision/remove-backgroundcURL

curl -X POST https://api.brainiall.com/v1/vision/remove-background \
  -H "Authorization: Bearer $BRAINIALL_KEY" \
  -F "image=@product_photo.jpg" \
  -F "format=png" \
  --output transparent_product.png

超解像画像拡大・アップスケール

POST/v1/vision/upscale

不自然なノイズを付加することなく、2倍・4倍・8倍の鮮明な輪郭拡大を実現。

2x, 4x, and 8x super-resolution scaling with artifact suppression
Restores sharp edges, textures, and typography without hallucinated noise

POST /v1/vision/upscalecURL

curl -X POST https://api.brainiall.com/v1/vision/upscale \
  -H "Authorization: Bearer $BRAINIALL_KEY" \
  -F "image=@lowres_banner.jpg" \
  -F "scale=4" \
  --output sharp_4k_banner.png

顔写真・ポートレート復元

POST/v1/vision/face-restore

不鮮明な顔写真の目元のディテールや肌の質感を自然に復元します。

Restores facial fidelity, eye iris clarity, and skin texture in blurry portraits
Preserves natural facial identity without artificial plastic smoothing

POST /v1/vision/face-restorecURL

curl -X POST https://api.brainiall.com/v1/vision/face-restore \
  -H "Authorization: Bearer $BRAINIALL_KEY" \
  -F "image=@archived_portrait.jpg" \
  --output restored_portrait.png

ビジュアルタギング・物体検出

POST/v1/vision/labels

マルチラベル分類、物体検出座標、シーン分類を構造化JSONで出力。

Multi-label classification, object detection coordinates, and scene categorization
Outputs structured JSON with bounding boxes and normalized confidence scores

POST /v1/vision/labelscURL

curl -X POST https://api.brainiall.com/v1/vision/labels \
  -H "Authorization: Bearer $BRAINIALL_KEY" \
  -F "image=@storefront.jpg"

# Response:
# {
#   "labels": [
#     { "name": "apparel", "confidence": 0.98, "box": [120, 80, 450, 600] },
#     { "name": "mannequin", "confidence": 0.94, "box": [100, 50, 480, 720] }
#   ]
# }