Object completion with stochastic completion fields predicts human behavior in recognizing degraded object drawings
Bibliographic record
Abstract
Biederman & Cooper (1991) showed that observers were better at classifying degraded line drawings of objects when shown only contour junctions than when shown only middle segments. We here provide an account of these results based on a low-level algorithm for object completion. The human visual system infers geometry not only for visible but also for occluded contours and surfaces. This provides a central theme to figural completion which lays out a computational framework to estimate paths that connect a set of boundary fragments. In our model, we use the Fokker-Planck Equation to extract a set of points with assigned orientations. These points represent the sources and sinks for a stochastic completion field (SCF) algorithm (Williams and Jacobs, 1995). The SCF algorithm produces a distribution of possible completion fields, where each field is a probability density function (PDF) that enables us to score each possible completion path. We tested the algorithm on the Snodgrass and Vanderwart (1980) dataset of 260 manually traced line drawings of objects. The traced objects were separated into two half images; one half with contour segments containing junctions, and the other with segments between junctions. We attempted to complete the half-drawings using the SCF algorithm. The completions replicated the original, intact drawings more faithfully for the half-drawings with junctions than those with middle segments. Our computational results show that contour completion is easier in objects with junctions than with middle segments, which aligns with Biederman & Cooper's behavioural result. Ultimately, the SCF is a method that is potentially useful for predicting which types of incomplete line drawings can be more easily completed by the human visual system. Moreover, SCF may form the computational basis for an image-computable implementation of the Good Continuation Gestalt grouping rule.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.001 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one teacher head, not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".