CoolFace
Apppublic

candy24/used-product-inspector

sourceHugging Faceotherupdated 1y agoView on Hugging Face
1likes
App README

Used Product Inspector (Gradio + Azure Custom Vision)

'Pre-listing Inspection for Used Goods'

  • —This AI simulation demo evaluates whether a used product can be listed on a secondhand marketplace, based on multiple real-angle images uploaded by the seller.
  • —It uses an Azure Custom Vision model to predict the Pre-Posting approval status as Yes, No, or Unclassified, simulating the internal inspection logic of secondhand trading platforms.

Powered by

  • —Azure Custom Vision
  • —Gradio UI

Purpose

'AI-powered pre-listing inspection logic'

  • —This system is designed from the perspective of secondhand marketplace platforms to maintain platform trust and protect buyers from potentially misleading listings or seller abuse.
  • —by identifying and filtering(screening) out ineligible products, pre rejecting through an AI-driven internal review process before posting.
  • —It simulates an AI-driven pre-listing validation process, using seller-taken from multiple angles real photos.
  • —To this screening system, determine seller-product eligibility status and then qualifies for posting.

How it Works

  1. 1.Sellers taken and upload multiple angle real photos of the product.
  2. 2.Thresholds for classification (Yes/No/Unclassified) are set.
  3. 3.The AI model analyzes the images.
  4. 4.The seller receives a result indicating eligibility and key reasons.

Notes on Unclassified Results

Images may be classified as "Unclassified" when:

  • —The photo is downloaded from the internet or copied from shopping websites.
  • —The image appears to be AI-generated or digitally synthesized.
  • —The product is too small, unclear, or only partially visible in the image.
  • —The AI cannot confidently detect the product due to poor image quality or ambiguous context.

In such cases, the product cannot be approved, as the image does not provide sufficient evidence of authenticity or product clarity.

Note: This model is intentionally designed for cases where sellers upload real-time photos they have taken themselves prior to posting items on second-hand platforms. It is not intended for evaluating stock photos or third-party content.

Summary

This demo serves as an AI simulation of a secondhand product inspection system. It predicts whether a used item is eligible for posting based on seller-uploaded, multi-angle real photos. The model aims to enhance platform safety, prevent fraud, and standardize listing eligibility decisions.

Predicted labels:

  • —Yes : Approved for Posting
  • —No : Rejected
  • —Unclassified : Undetermined — likely due to image issues

The system returns the Posting eligibility status of the product as Yes, No, or Unclassified.

Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference