README.md

metadata

title: Amazon Trailer Inspector
emoji: 🚛
colorFrom: blue
colorTo: indigo
sdk: docker
app_file: app.py
pinned: false
license: mit

🚛 Amazon Trailer Inspector — REST API

AI-powered trailer inspection pipeline exposed as a REST API.
Submit up to 6 labeled images and receive a structured JSON inspection report.

---

Endpoint

POST /inspect

Submit all images in one request. All aspects are analyzed in parallel.

Request Body (JSON)

{
  "images": [
    { "label": "front_left",  "image_base64": "<base64-string>" },
    { "label": "front_right", "image_base64": "<base64-string>" },
    { "label": "rear_left",   "image_base64": "<base64-string>" },
    { "label": "rear_right",  "image_base64": "<base64-string>" },
    { "label": "inside",      "image_base64": "<base64-string>" },
    { "label": "door",        "image_base64": "<base64-string>" }
  ]
}

Accepted labels: front_left, front_right, rear_left, rear_right, inside, door

• Images can be raw base64 OR a data-URI (data:image/jpeg;base64,...)
• All 6 images are optional — only send the ones you have
• Each label can appear at most once per request
• rear_left and rear_right must both be present for full rear confirmation

---

Response Body (JSON)

{
  "status": "success",
  "images_received": ["front_left", "front_right", "rear_left", "rear_right", "inside", "door"],
  "labels_missing": [],
  "report": {
    "front": {
      "label": "Front Left / Right",
      "images_provided": ["front_left", "front_right"],
      "components": {
        "Sensors":          "detected",
        "GPS Device":       "missing",
        "Prime Logo":       "detected",
        "Trailer ID Label": "detected"
      },
      "notes": []
    },
    "rear": {
      "label": "Rear Left / Right",
      "images_provided": ["rear_left", "rear_right"],
      "components": {
        "Side Skirts / Fins": { "status": "detected",                       "count": "2/2" },
        "Edge Kit":           { "status": "partially detected (left side only)", "count": "1/2" }
      },
      "notes": []
    },
    "inside": {
      "label": "Inside Trailer",
      "images_provided": ["inside"],
      "components": {
        "Side Guards": "detected",
        "Flooring":    "detected"
      },
      "notes": []
    },
    "door": {
      "label": "Door Details",
      "images_provided": ["door"],
      "components": {
        "Latch Kit & Lash Links": "detected",
        "Grote LED Lights":       "missing"
      },
      "notes": []
    }
  }
}

---

Component Values

Aspect	Component Values
Front	"detected" or "missing" (OR logic — detected if found in either left or right image)
Rear	"detected", "partially detected (left side only)", "partially detected (right side only)", or "missing" — plus count ("2/2", "1/2", "0/2")
Inside	"detected" or "missing"
Door	"detected" or "missing"

---

Notes Field

The notes array in each aspect section contains warnings such as:

• "front_left: image missing from input" — label was not submitted
• "rear_right: invalid image (wrong angle/side)" — model rejected the image
• "door: invalid image (not a valid rear door view)" — wrong image type

---

Other Endpoints

Method	Path	Description
GET	/	Service info
GET	/health	HF_TOKEN status + models
GET	/docs	Interactive Swagger UI

---

Setup

Space Settings → Repository Secrets → add GEMINI_API_KEY

Get a free Gemini API key at aistudio.google.com/apikey
No credit card required — free tier includes 1,500 requests/day.

---

Models (Google Gemini, free tier)

1. gemini-2.0-flash — primary (best quality + speed, free)
2. gemini-2.0-flash-lite — fallback 1 (lighter 2.0 variant, free tier)
3. gemini-1.5-flash-latest — fallback 2 (1.5 gen, stable free tier)