CITATION.cff

cff-version: 1.2.0
message: >-
  If you use ZPE Video in your work, please cite this software record.
type: software
title: "ZPE Video: Perception Receipts for AI Video Pipelines"
authors:
  - name: Zer0pa (Pty) Ltd
    email: architects@zer0pa.ai
version: 0.1.0
date-released: "2026-04-24"
repository-code: https://github.com/Zer0pa/ZPE-Video
url: https://github.com/Zer0pa/ZPE-Video
license: LicenseRef-Zer0pa-SAL-7.1
abstract: >-
  ZPE Video is a compact, deterministic binary format for recording what a
  detector and tracker saw in a video. Under default settings, every
  conforming writer in any language emits byte-identical output, making the
  SHA-256 of a receipt a stable cryptographic reference to what the model
  saw. The core library is pure stdlib; the wire format is implementable in
  under 100 lines of any language with stdlib struct, zlib, and crc32
  primitives. Applications include C2PA-style AI-perception credentials,
  regulated chain-of-custody for detector output, video-LLM object-memory
  caches with integrity requirements, and cross-platform
  training-data provenance for vision pipelines.
keywords:
  - video
  - perception
  - receipts
  - provenance
  - C2PA
  - chain-of-custody
  - video-LLM
  - object-memory
  - deterministic-binary
  - cross-writer-hash-stability
  - zero-point encoding
preferred-citation:
  type: article
  title: ZPE Architecture Originating Publication (Defensive Prior-Art Disclosure)
  authors:
    - name: Zer0pa (Pty) Ltd
      email: architects@zer0pa.ai
  year: 2026
  url: https://github.com/Zer0pa/ZPE-Video
identifiers:
  - type: url
    value: https://github.com/Zer0pa/ZPE-Video
    description: Live repository