Question

To what extent can visual features alone predict the approximate historical period and cultural-geographic region of an artwork, and how well do these predictions generalise to artworks, artists and collections not encountered during training?

The question is experimental. That a model produces a prediction says nothing about whether the prediction is informative; Artifex is built to measure that, and to show the answer to users.

Scope

• Objects: paintings and painted imagery (including illustrated manuscript folios and album leaves), as catalogued by The Metropolitan Museum of Art and The Cleveland Museum of Art and released under CC0.
• Targets: a broad cultural-geographic region (with narrower traditions where data allow) and an ordered date bin.
• Input: the image only. Titles, artist names, inscriptions read by people, and other text are not used for prediction. Text search exists in the collection browser purely for browsing.

Hypotheses

H1 — Region is recoverable from appearance well above chance for traditions that are well represented in the reference collection, because format, materials, pictorial conventions and support (e.g. hanging scroll vs. canvas) differ systematically between traditions.

H2 — Period is harder than region, and errors are mostly between adjacent bins: neighbouring centuries share conventions, and within-tradition change is gradual. Ordinal metrics (within-one-bin accuracy, error in years) should therefore be much better than exact-bin accuracy.

H3 — Performance drops for unseen artists and more so for an unseen institution. Artist-disjoint splits remove style memorisation; collection-disjoint tests additionally change photography, colour processing and collecting history. The size of the drop indicates how much of the signal is collection-specific.

H4 — Part of the signal is low-level (palette, varnish yellowing, paper tone, photographic conditions). A colour/texture baseline will beat the majority class, and greyscale input will reduce performance — but DINOv2 should outperform the low-level baseline if higher-level structure matters.

H5 — Agreement between the classifier and nearest-neighbour retrieval is a useful reliability signal: predictions should be more accurate when the two agree.

H6 — Calibration is necessary. Raw logistic-regression scores will be miscalibrated with balanced class weights; temperature scaling on validation data should reduce calibration error on test data drawn from the same collections, but cannot be relied upon for inputs from other distributions.

What would count as evidence

Each hypothesis is assessed against baselines run by the same code on the same splits (majority class, k-NN, random retrieval, ablation variants). The evaluation report of each run contains the numbers; this document deliberately contains none. Confirmations and refutations are recorded in the run's report.md and the Experiments page.

Relevance

Computational art history has studied style, influence and dating with learned representations (e.g. work on WikiArt style classification, the Replica project on visual pattern discovery, and large-scale analyses of museum collections). Artifex focuses on a narrower, practical setting — what a single photograph can tell a non-specialist about an unfamiliar painting — and on the honesty of that answer: abstention, calibration, provenance and leakage-aware evaluation.