Report examples
Read img2det results as evidence, not a one-word verdict.
These planned public-safe cases show the review language that will accompany reproducible reports. They are not completed test results; each remains blocked until its actual input hash and report are published under the benchmark-v1 release checklist release gate.
Camera metadata baseline
Planned condition: Camera original: Before publication, run the named input through the production evidence checker without changing the documented input recipe.
Evidence to inspect: Compare the report’s strongest evidence category with the documented fixture condition.
Interpretation: Publication is blocked until a real input hash, dated report JSON, and human review note are recorded.
Limit: This is a planned controlled case, not a completed benchmark result and not a measure of real-world prevalence or detector accuracy.
Camera metadata removed
Planned condition: Camera original: Before publication, run the named input through the production evidence checker without changing the documented input recipe.
Evidence to inspect: Compare the report’s strongest evidence category with the documented fixture condition.
Interpretation: Publication is blocked until a real input hash, dated report JSON, and human review note are recorded.
Limit: This is a planned controlled case, not a completed benchmark result and not a measure of real-world prevalence or detector accuracy.
Camera image recompressed once
Planned condition: Camera original: Before publication, run the named input through the production evidence checker without changing the documented input recipe.
Evidence to inspect: Compare the report’s strongest evidence category with the documented fixture condition.
Interpretation: Publication is blocked until a real input hash, dated report JSON, and human review note are recorded.
Limit: This is a planned controlled case, not a completed benchmark result and not a measure of real-world prevalence or detector accuracy.
Camera image converted to WebP
Planned condition: Camera original: Before publication, run the named input through the production evidence checker without changing the documented input recipe.
Evidence to inspect: Compare the report’s strongest evidence category with the documented fixture condition.
Interpretation: Publication is blocked until a real input hash, dated report JSON, and human review note are recorded.
Limit: This is a planned controlled case, not a completed benchmark result and not a measure of real-world prevalence or detector accuracy.
AI provenance-marker baseline
Planned condition: AI original export: Before publication, run the named input through the production evidence checker without changing the documented input recipe.
Evidence to inspect: Compare the report’s strongest evidence category with the documented fixture condition.
Interpretation: Publication is blocked until a real input hash, dated report JSON, and human review note are recorded.
Limit: This is a planned controlled case, not a completed benchmark result and not a measure of real-world prevalence or detector accuracy.
AI marker with metadata retained
Planned condition: AI original export: Before publication, run the named input through the production evidence checker without changing the documented input recipe.
Evidence to inspect: Compare the report’s strongest evidence category with the documented fixture condition.
Interpretation: Publication is blocked until a real input hash, dated report JSON, and human review note are recorded.
Limit: This is a planned controlled case, not a completed benchmark result and not a measure of real-world prevalence or detector accuracy.
Use the examples with the live checker
Run the file first, then compare the report wording with the closest example. If the evidence is marker-only, missing, or platform-stripped, ask for the original file before making stronger claims.