How estimates are formed
AI Card Grading Methodology
A hybrid image and AI pipeline
1. Image preparation
The system detects the card area, creates a reviewed crop, and generates full-card and diagnostic views. The production server can use an OpenCV service, with a Sharp-based processing path available as a fallback.
2. Centering evidence
Computer-vision measurements attempt to locate card and border geometry. Low-confidence or unsuitable geometry is not treated as authoritative measured centering.
3. Focused condition passes
Separate AI passes assess visible centering, corner, edge, and surface evidence rather than asking one response to infer every dimension at once.
4. Consistent assembly
The default production path assembles focused results locally, normalises grade math, and records which dimensions were defaulted instead of measured.
What the report means
The report provides condition findings and estimates expressed in formats familiar to PSA, BGS, and TAG collectors. Those outputs are CardRevive's estimates. They are not copies of a grading company's proprietary process, and the TAG-style figure is not a TAG photometric result.
Grade ranges communicate uncertainty better than a single number. Qualitative confidence notes describe the evidence available to the analysis; they are not a statistically calibrated probability of matching a professional grade.
Known methodological limits
- A flat image cannot reproduce touch, magnification, angled light, spectroscopy, or an in-hand inspection.
- Glare, scanner shadows, plastic, compression, focus, and colour processing can hide real defects or create false ones.
- The service does not authenticate cards or reliably detect counterfeits, trimming, recolouring, restoration, or every alteration.
- Professional grading standards, graders, and submission policies can vary and change independently of CardRevive.
- Borderless, die-cut, textured, transparent, and unusually shaped cards can reduce the reliability of geometric measurements.
AI estimates and CardRevive submission services are separate
This AI report estimates PSA, BGS, and a TAG-style condition result. Separately, CardRevive operates as an independent grading middleman for PSA, BGS, PCG, ACE, and CGC. The grading house, not CardRevive, makes the official authentication and grade decision.
PCG, ACE, and CGC do not receive house-specific grade estimates in this report. TAG is an estimate-only reference and is not in CardRevive's currently published middleman service list. CardRevive's starting prices effective 11 July 2026 are:
- PSA: from A$172 AUD per card
- BGS: from A$55 AUD per card
- PCG: from A$30 AUD per card (CardRevive exclusive)
- ACE: from A$60 AUD per card
- CGC: from A$40 AUD per card
These are starting prices, not guaranteed totals. Final service fit, fees, timing, declared-value treatment, shipping, and any upcharge are confirmed before dispatch. View the current CardRevive grading details or start a CardRevive submission.
How returned grades become accuracy data
An owner can record a later PSA, BGS, or TAG result and optionally supply slab-label evidence. A user-entered result remains visibly unverified until CardRevive reviews its certificate or evidence. It enters public aggregate calculations only after review and separate anonymous-aggregate consent; duplicate certificate fingerprints, rejected evidence, impossible grades, and records without an earlier company-specific estimate are excluded.
The accuracy page defines exact-grade and within-one-grade calculations, publishes the denominator beside every percentage, and withholds percentages when the verified sample is too small. It does not guarantee a maximum difference from a future professional grade.
If you are deciding whether to pay a submission fee, inspect the card yourself, read the current grading company rules, and treat this report as one input. The scan guide can improve the evidence available, not guarantee the outcome.