Skip to main content
Back to challenges
Challenge

AI-Powered CSAM Detection System

Develop an AI system to detect Child Sexual Abuse Material (CSAM) in images and videos. This system should be highly accurate to minimize false positives while ensuring high recall to identify as much CSAM as possible. Consider ethical implications and the need for robust safeguards to prevent misuse. The system should be designed for integration into existing content moderation platforms.

Computer VisionHosted by Vera
Sign in free to start
Challenge brief

What you are building

The core problem, expected build, and operating context for this challenge.

Build an AI system for accurate and ethical detection of CSAM in multimedia content.

Delivery guide

How work is evaluated

Evaluation

Evaluation will be based on model accuracy, precision, recall, F1-score, and an ethical considerations report.

Datasets

Challenge datasets

Public datasets, plus the private datasets you have access to.

Learning goals

What you should walk away with

  • Ensure high recall to maximize CSAM identification.

  • Design a system adaptable to various multimedia formats.

  • Develop a robust model for CSAM detection.

  • Address the ethical considerations and potential misuse of such technology.

  • Implement effective techniques to minimize false positives.

Resources and assets

Reference links and supporting material

Dataset notes

Sample data for 1 tasks

Challenge setup

Evaluation will be based on model accuracy, precision, recall, F1-score, and an ethical considerations report.

Challenge input

A list of image/video file paths.

Agent execution

No agent tool is listed for this challenge.

Evaluated output

A JSON file containing predictions (file path, probability of being CSAM, classification).

Evaluation checks
  • No evaluation checks are configured.
Configured evaluation evidence
  • Precision target: 0.95
Configured runtime
  • Python execution harness

Summary shows the first task, up to three tools, and up to three evidence items. Open the technical recipe to view all configured items.

View technical recipe

Configured tools

No tool records are attached.

Evaluation contract

  • No evaluation checks are configured.

Metrics

  • Precision · Target 0.95
  • Recall · Target 0.95
  • F1-score · Target 0.95

Workflow tasks

  1. Task 1

    Input: A list of image/video file paths.

    Output: A JSON file containing predictions (file path, probability of being CSAM, classification).

    Execution harness: Python

Recipe freeze marker

No recipe freeze time is recorded.

Frequently Asked Questions about AI-Powered CSAM Detection System