Multiple object tracking is scene-based, not image-based
Bibliographic record
Abstract
Multiple object tracking (MOT) is the ability to individuate a moving object based solely on its spatial-temporal history. We examined whether MOT is based on a scene-based (allocentric) or image-based (egocentric) representation. Observers viewed 16 objects moving in a depicted 3D wireframe box. On each trial, 2, 4, or 6 objects were briefly tagged as the ‘target’ class. All objects then underwent 10 s of random motion (1 or 6 deg/s) before stopping. A single object was then tagged, which the observer identified as a target or a non-target. Preliminary experiments established that MOT was impaired by increases in both size of the target class and speed of object motion. Next, the motion pattern of the 3D box was manipulated. Thus, in addition to varying the speed of objects relative to the center of the box (object motion), the motion of the whole box was varied (scene motion). Unlike variations in object motion, which had a large influence on accuracy, variations in scene motion had no measurable influence. This was true whether the scene underwent translation, zoom, rotation, or even a combination of all three motions (‘combined motion’). To tax the ability to use a scene-based representation, we projected the ‘combined motion’ condition onto an obliquely viewed surface. This created retinal motions of the objects and box consistent with an orthogonal view, but the apparent motions underwent large changes because of the affine stretching of the projected image. Nonetheless, MOT accuracy was unaffected. Accuracy was only reduced when we projected the ‘combined motion’ onto a convex corner formed from the junction of two surfaces, the same conditions under which pictorial shape constancy is no longer possible. These results imply that MOT is accomplished with a scene-based representation. It is motion of objects relative to the larger scene that determines performance, not motion of objects relative to egocentric landmarks like retinal location.
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 machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.004 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.000 |
| Science and technology studies | 0.000 | 0.001 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.000 | 0.001 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.001 | 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 source (direct Gemma or distilled Codex), 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".