😊 Emotion Recognition

The /emotion endpoint analyzes faces in an image and detects the dominant emotions expressed by each individual. It is ideal for building apps that react to user sentiment, conduct surveys, or monitor engagement.

---

📥 Endpoint : POST /emotion

Base URL: https://api.visionary.ai/v1/emotion

---

🔐 Headers

Key	Value	Required
Authorization	Bearer YOUR_API_KEY	✅
Content-Type	application/json or multipart/form-data	✅

---

📸 Request Body

Option 1 – JSON (Base64)

{
  "image": "data:image/png;base64,iVBORw0KGgoAAAANSUhEUg..."
}

Option 2 – Multipart (recommended)

Use image as the field name when uploading.

---

🧪 Example (curl – multipart)

curl -X POST https://api.visionary.ai/v1/emotion \
  -H "Authorization: Bearer demo-api-key" \
  -F "image=@/path/to/image.jpg"

---

✅ Response

{
  "faces": [
    {
      "bounding_box": {
        "x": 120,
        "y": 80,
        "width": 140,
        "height": 140
      },
      "emotions": [
        {
          "label": "happy",
          "confidence": 0.92
        },
        {
          "label": "neutral",
          "confidence": 0.07
        }
      ]
    }
  ]
}

🧠 Fields

Field	Type	Description
bounding_box	object	Location of detected face (x, y, width, height)
emotions	array	List of probable emotions and their confidence
label	string	Detected emotion (e.g. happy, sad, angry)
confidence	float	Value between 0 and 1

---

😊 Possible Emotions

• happy

• sad

• angry

• surprised

• fearful

• disgusted

• neutral

---

🚫 Error Responses

Code	Message	Cause
400	"No face detected"	No recognizable face in image
401	"Unauthorized"	Missing or invalid API key
422	"Unprocessable Entity"	Bad image format or corrupted

---

📌 Notes

• Works best with frontal or slightly angled faces.

• Can detect multiple faces in the same image.

• Results are approximate and may vary based on image quality.

---

🔗 Next endpoint

➡️ Try the /nsfw-filter endpoint to moderate inappropriate content.