phenomenon · claim-type · cross-domain-transfer · mechanism · claim-relation · replication-status · limitation — grounded in the full text| phenomenon | a field whose measure of success is a threshold and whose object vocabulary is a benchmark's class list - 20, 200 or 91 depending which benchmark— “In practice, current detectors focus mainly on structured object categories, such as the 20, 200 and 91 object classes in PASCAL VOC (Everingham et al. 2010), ILSVRC (Russakovsky et al. 2015) and MS COCO (Lin et al. 2014) respectively. Clearly, the number of object categories under consideration in existing benchmark datasets is much smaller… | |
| claim-type | critique— “In practice, current detectors focus mainly on structured object categories, such as the 20, 200 and 91 object classes in PASCAL VOC (Everingham et al. 2010), ILSVRC (Russakovsky et al. 2015) and MS COCO (Lin et al. 2014) respectively. Clearly, the number of object categories under consideration in existing benchmark datasets is much smaller… | |
| method — evidence | — “More than 300 research contributions are included in this survey, covering many aspects of generic object detection: detection frameworks, object feature representation, object proposal generation, context modeling, training strategies, and evaluation metrics. We finish the survey by identifying promising directions for future research.” | |
| object of study — evidence | — “The goal of object detection is to determine whether there are any instances of objects from given categories (such as humans, cars, bicycles, dogs or cats) in an image and, if present, to return the spatial location and extent of each object instance” | |
| research question — evidence | — “More than 300 research contributions are included in this survey, covering many aspects of generic object detection: detection frameworks, object feature representation, object proposal generation, context modeling, training strategies, and evaluation metrics. We finish the survey by identifying promising directions for future research.” | |
| contribution — evidence | — “More than 300 research contributions are included in this survey, covering many aspects of generic object detection: detection frameworks, object feature representation, object proposal generation, context modeling, training strategies, and evaluation metrics. We finish the survey by identifying promising directions for future research.” | |
| transfer | from information retrieval: precision and recall from information retrieval, carried into vision as the basis of average precision and then of the field's single headline number— “The most commonly used metric is Average Precision (AP), derived from precision and recall. AP is usually evaluated in a category specific manner, i.e., computed for each object category separately. To compare performance over all object categories, the mean AP (mAP) averaged over all object categories is adopted as the final measure of… | transfers[0] |
| mechanism | correct is a threshold a detection counts as a true positive when its category matches AND its overlap with the ground-truth box is not smaller than a predefined threshold. Every precision, recall and AP number in the field is downstream of that number - and the number is a convention, not a derivation— “A predicted detection (b, c, p) is regarded as a True Positive (TP) if • The predicted category c equals the ground truth label cg. • The overlap ratio IOU (Intersection Over Union) (Everingham et al. 2010; Russakovsky et al. 2015) IOU(b, bg) = area (b ∩ bg) area (b ∪ bg) , (4) between the predicted BB b and the ground truth bg is not smaller… | mechanisms[0] |
| mechanism | the same metric name, two quantities PASCAL VOC's mAP is computed at a single IoU of 0.5; MS COCO's is averaged across 0.5 to 0.95 in steps of 0.05. Both are called mAP. A survey of the field has to print a table of which benchmark means which, and any comparison across that boundary is comparing two definitions— “Since the introduction of MS COCO, more attention has been placed on the accuracy of the bounding box location. Instead of using a fixed IOU threshold, MS COCO introduces a few metrics (summarized in Table 5) for characterizing the performance of an object detector. For instance, in contrast to the traditional mAP computed at a single IoU of… | mechanisms[1] |
| mechanism | a metric subdivided because the dataset was skewed 41% of MS COCO objects are small and 24% are large, so size-stratified metrics were introduced. THE AGGREGATE WAS MISLEADING BECAUSE OF WHAT THE DATASET CONTAINED, and the response was to change the measure rather than the data— “Because 41% of the objects in MS COCO are small and 24% are large, metrics A P small coco are also introduced.” | mechanisms[2] |
| mechanism | the object vocabulary is a benchmark's class list 20 classes in PASCAL VOC, 200 in ILSVRC, 91 in MS COCO. `Object detection` means detection of whichever of those lists the paper trained against, and the survey says plainly that all of them are much smaller than what a human recognises— “In practice, current detectors focus mainly on structured object categories, such as the 20, 200 and 91 object classes in PASCAL VOC (Everingham et al. 2010), ILSVRC (Russakovsky et al. 2015) and MS COCO (Lin et al. 2014) respectively. Clearly, the number of object categories under consideration in existing benchmark datasets is much smaller… | mechanisms[3] |
| mechanism | appearance does not determine the label, in either direction one figure panel shows a single class spanning wildly different images; another shows four near-identical images from four different classes. So a detector cannot be evaluated on visual similarity, and the annotation is the only ground truth there is— “Changes in appearance of the same class with variations in imaging conditions (a–h). There is an astonishing variation in what is meant to be a single object class (i). In contrast, the four images in j appear very similar, but in fact are from four different object classes.” | mechanisms[4] |
| mechanism | THE LONGEST TERM LIST IN THE CORPUS, AND IT IS NOT A LIST OF SYNONYMS - IT IS A LIST OF TERMS WHOSE BOUNDARIES DISAGREE BETWEEN PAPERS, AND IT SITS IN A FOOTNOTE: `To the best of our knowledge, there is no universal agreement in the literature on the definitions of various vision subtasks. Terms such as detection, localization, recognition, classification, categorization, verification, identification, annotation, labeling, and understanding are often differently defined (Andreopoulos and Tsotsos 2013)`. Ten terms, attributed, declared unagreed, in a footnote - and the survey then has to pick.— “To the best of our knowledge, there is no universal agreement in the literature on the definitions of various vision subtasks. Terms such as detection, localization, recognition, classification, categorization, verification, identification, annotation, labeling, and understanding are often differently defined (Andreopoulos and Tsotsos… | mechanisms[5] |
| critiques | the benchmark class lists themselves - 20, 200 and 91 categories against a real world of 10^4 to 10^5— “In practice, current detectors focus mainly on structured object categories, such as the 20, 200 and 91 object classes in PASCAL VOC (Everingham et al. 2010), ILSVRC (Russakovsky et al. 2015) and MS COCO (Lin et al. 2014) respectively. Clearly, the number of object categories under consideration in existing benchmark datasets is much smaller… | claim_relations[0] |
| qualifies | mAP as a comparable number, by showing that PASCAL VOC and MS COCO compute it at different IoU regimes— “Since the introduction of MS COCO, more attention has been placed on the accuracy of the bounding box location. Instead of using a fixed IOU threshold, MS COCO introduces a few metrics (summarized in Table 5) for characterizing the performance of an object detector. For instance, in contrast to the traditional mAP computed at a single IoU of… | claim_relations[1] |
| builds-on | more than 300 research contributions, organised by component of the detection pipeline— “More than 300 research contributions are included in this survey, covering many aspects of generic object detection: detection frameworks, object feature representation, object proposal generation, context modeling, training strategies, and evaluation metrics. We finish the survey by identifying promising directions for future research.” | claim_relations[2] |
| distinguishes-from | detection of specific instances, which the survey separates out as essentially a matching problem— “As illustrated in Fig. 2, object detection can be grouped into one of two types (Grauman and Leibe 2011; Zhang et al. 2013): detection of specific instances versus the detection of broad categories. The first type aims to detect instances of a particular object (such as Donald Trump’s face, the Eiffel Tower, or a neighbor’s dog), essentially… | claim_relations[3] |
| sense | a correct detection - AS DEFINED: the predicted category equals the ground truth label AND the intersection-over-union with the ground-truth box is not smaller than a predefined threshold— “A predicted detection (b, c, p) is regarded as a True Positive (TP) if • The predicted category c equals the ground truth label cg. • The overlap ratio IOU (Intersection Over Union) (Everingham et al. 2010; Russakovsky et al. 2015) IOU(b, bg) = area (b ∩ bg) area (b ∪ bg) , (4) between the predicted BB b and the ground truth bg is not smaller… | senses[0] |
| sense | mAP - UNDER PASCAL VOC: computed at a single IoU of 0.5— “Since the introduction of MS COCO, more attention has been placed on the accuracy of the bounding box location. Instead of using a fixed IOU threshold, MS COCO introduces a few metrics (summarized in Table 5) for characterizing the performance of an object detector. For instance, in contrast to the traditional mAP computed at a single IoU of… | senses[1] |
| sense | mAP - UNDER MS COCO: averaged across IoU values from 0.5 to 0.95 in steps of 0.05, plus separate metrics for small, medium and large objects— “Since the introduction of MS COCO, more attention has been placed on the accuracy of the bounding box location. Instead of using a fixed IOU threshold, MS COCO introduces a few metrics (summarized in Table 5) for characterizing the performance of an object detector. For instance, in contrast to the traditional mAP computed at a single IoU of… | senses[2] |
| sense | object detection - OF SPECIFIC INSTANCES: matching a particular object, such as one face or one tower— “As illustrated in Fig. 2, object detection can be grouped into one of two types (Grauman and Leibe 2011; Zhang et al. 2013): detection of specific instances versus the detection of broad categories. The first type aims to detect instances of a particular object (such as Donald Trump’s face, the Eiffel Tower, or a neighbor’s dog), essentially… | senses[3] |
| sense | object detection - OF BROAD CATEGORIES: the generic problem this survey is about, whose category list is whatever the benchmark defines— “In practice, current detectors focus mainly on structured object categories, such as the 20, 200 and 91 object classes in PASCAL VOC (Everingham et al. 2010), ILSVRC (Russakovsky et al. 2015) and MS COCO (Lin et al. 2014) respectively. Clearly, the number of object categories under consideration in existing benchmark datasets is much smaller… | senses[4] |
| sense | an object class - AS A VISUAL GROUPING: what the four similar-looking images in Fig. 6j would suggest, and which the caption says is wrong— “Changes in appearance of the same class with variations in imaging conditions (a–h). There is an astonishing variation in what is meant to be a single object class (i). In contrast, the four images in j appear very similar, but in fact are from four different object classes.” | senses[5] |
| sense | an object class - AS A LABEL: a benchmark's annotation, which can hold an astonishing variation in appearance and can separate images that look alike— “Changes in appearance of the same class with variations in imaging conditions (a–h). There is an astonishing variation in what is meant to be a single object class (i). In contrast, the four images in j appear very similar, but in fact are from four different object classes.” | senses[6] |
| sense | TEN NAMES FOR OVERLAPPING TASKS, AND THE FIELD IS DECLARED TO HAVE NO AGREEMENT ABOUT ANY OF THEM: `there is no universal agreement in the literature on the definitions of various vision subtasks. Terms such as detection, localization, recognition, classification, categorization, verification, identification, annotation, labeling, and understanding are often differently defined`. A survey cannot proceed without saying which it means, and the field supplies ten words and no map.— “To the best of our knowledge, there is no universal agreement in the literature on the definitions of various vision subtasks. Terms such as detection, localization, recognition, classification, categorization, verification, identification, annotation, labeling, and understanding are often differently defined (Andreopoulos and Tsotsos… | senses[7] |
| control | MS COCO's average over ten IoU thresholds instead of a single 0.5 — held— “Since the introduction of MS COCO, more attention has been placed on the accuracy of the bounding box location. Instead of using a fixed IOU threshold, MS COCO introduces a few metrics (summarized in Table 5) for characterizing the performance of an object detector. For instance, in contrast to the traditional mAP computed at a single IoU of… — rules out a detector tuned to clear one overlap value - averaging across 0.5 to 0.95 prices localisation accuracy into the score— “Since the introduction of MS COCO, more attention has been placed on the accuracy of the bounding box location. Instead of using a fixed IOU threshold, MS COCO introduces a few metrics (summarized in Table 5) for characterizing the performance of an object detector. For instance, in contrast to the traditional mAP computed at a single IoU of… | controls[0] |
| control | size-stratified metrics introduced alongside the aggregate, after the object size distribution of the dataset was measured — held— “Because 41% of the objects in MS COCO are small and 24% are large, metrics A P small coco are also introduced.” — rules out a headline number dominated by the size class that happens to be most common - 41% of MS COCO objects are small— “Because 41% of the objects in MS COCO are small and 24% are large, metrics A P small coco are also introduced.” | controls[1] |
| control | the challenge's two rankings compared - by mAP and by number of object categories won - before adopting mAP as the team measure — held— “In object detection challenges, such as PASCAL VOC and ILSVRC, the winning entry of each object category is that with the highest AP score, and the winner of the challenge is the team that wins on the most object categories. The mAP is also used as the measure of a team’s Footnote 3 continued performance, and is justified since the ranking of… — rules out an aggregate standing in for per-category performance without evidence; the justification is that the two rankings were always the same— “In object detection challenges, such as PASCAL VOC and ILSVRC, the winning entry of each object category is that with the highest AP score, and the winner of the challenge is the team that wins on the most object categories. The mAP is also used as the measure of a team’s Footnote 3 continued performance, and is justified since the ranking of… | controls[2] |
| control | AP evaluated in a category-specific manner and only then averaged, rather than pooled across categories — held— “The most commonly used metric is Average Precision (AP), derived from precision and recall. AP is usually evaluated in a category specific manner, i.e., computed for each object category separately. To compare performance over all object categories, the mean AP (mAP) averaged over all object categories is adopted as the final measure of… — rules out a common category's performance swamping a rare one in the aggregate— “The most commonly used metric is Average Precision (AP), derived from precision and recall. AP is usually evaluated in a category specific manner, i.e., computed for each object category separately. To compare performance over all object categories, the mean AP (mAP) averaged over all object categories is adopted as the final measure of… | controls[3] |
| control | figure panels showing one class across wildly different appearances and four classes across near-identical appearances — absent-declared-unreported— “Changes in appearance of the same class with variations in imaging conditions (a–h). There is an astonishing variation in what is meant to be a single object class (i). In contrast, the four images in j appear very similar, but in fact are from four different object classes.” | controls[4] |
| weight | the benchmark class lists themselves - 20, 200 and 91 categories against a real world of 10^4 to 10^5 — 0.95— “In practice, current detectors focus mainly on structured object categories, such as the 20, 200 and 91 object classes in PASCAL VOC (Everingham et al. 2010), ILSVRC (Russakovsky et al. 2015) and MS COCO (Lin et al. 2014) respectively. Clearly, the number of object categories under consideration in existing benchmark datasets is much smaller… | claim_relations[0] |
| weight | mAP as a comparable number, by showing that PASCAL VOC and MS COCO compute it at different IoU regimes — 0.95— “Since the introduction of MS COCO, more attention has been placed on the accuracy of the bounding box location. Instead of using a fixed IOU threshold, MS COCO introduces a few metrics (summarized in Table 5) for characterizing the performance of an object detector. For instance, in contrast to the traditional mAP computed at a single IoU of… | claim_relations[1] |
| weight | more than 300 research contributions, organised by component of the detection pipeline — 0.9— “More than 300 research contributions are included in this survey, covering many aspects of generic object detection: detection frameworks, object feature representation, object proposal generation, context modeling, training strategies, and evaluation metrics. We finish the survey by identifying promising directions for future research.” | claim_relations[2] |
| weight | detection of specific instances, which the survey separates out as essentially a matching problem — 0.9— “As illustrated in Fig. 2, object detection can be grouped into one of two types (Grauman and Leibe 2011; Zhang et al. 2013): detection of specific instances versus the detection of broad categories. The first type aims to detect instances of a particular object (such as Donald Trump’s face, the Eiffel Tower, or a neighbor’s dog), essentially… | claim_relations[3] |
| audit | appearance —distinct-from→ object class — Audit 2026-09-28: DOUBTFUL - read against its span and kept, but the span contrasts intra-class variation with inter-class similarity; distinct-from is loose.— “Changes in appearance of the same class with variations in imaging conditions (a–h). There is an astonishing variation in what is meant to be a single object class (i). In contrast, the four images in j appear very similar, but in fact are from four different object classes.” | concept_edges[1] |
| differs in (stated by the tagger, no quotation) | appearance —distinct-from→ object class — whether two images look alike or carry the same label | concept_edges[1] |
| relation | appearance —distinct-from→ object class — one class holds an astonishing variation in appearance, and four near-identical images belong to four different classes - the label is not recoverable from how things look, in either direction— “Changes in appearance of the same class with variations in imaging conditions (a–h). There is an astonishing variation in what is meant to be a single object class (i). In contrast, the four images in j appear very similar, but in fact are from four different object classes.”— “There is an astonishing variation in what is meant to be a single object class (i). In contrast, the four images in j appear very similar, but in fact are from four different object classes.” | concept_edges[1] |
| weight | appearance —distinct-from→ object class — 0.95— “Changes in appearance of the same class with variations in imaging conditions (a–h). There is an astonishing variation in what is meant to be a single object class (i). In contrast, the four images in j appear very similar, but in fact are from four different object classes.” | concept_edges[1] |
| audit | intraclass variations —distinct-from→ object categories — Audit 2026-09-28: DOUBTFUL - read against its full span and kept, but the span adds the number of object categories to intraclass variations as a second challenge; distinct-from is loose.— “In addition to intraclass variations, the large number of object categories, on the order of 104–105, demands great discrimination power from the detector to distinguish between subtly different interclass variations, as illustrated in Fig. 6j.” | concept_edges[2] |
| differs in (stated by the tagger, no quotation) | intraclass variations —distinct-from→ object categories — whether the difficulty is within one label or between labels | concept_edges[2] |
| relation | intraclass variations —distinct-from→ object categories — the two named sources of detection difficulty - variation among instances sharing a label, and the sheer number of labels to be told apart— “In addition to intraclass variations, the large number of object categories, on the order of 104–105, demands great discrimination power from the detector to distinguish between subtly different interclass variations, as illustrated in Fig. 6j.” | concept_edges[2] |
| weight | intraclass variations —distinct-from→ object categories — 0.95— “In addition to intraclass variations, the large number of object categories, on the order of 104–105, demands great discrimination power from the detector to distinguish between subtly different interclass variations, as illustrated in Fig. 6j.” | concept_edges[2] |
| differs in (stated by the tagger, no quotation) | specific instances —distinct-from→ broad categories — whether the target is one object or a class of them | concept_edges[3] |
| relation | specific instances —distinct-from→ broad categories — detecting one particular object is essentially a matching problem; detecting a category is the generic task this survey is about— “As illustrated in Fig. 2, object detection can be grouped into one of two types (Grauman and Leibe 2011; Zhang et al. 2013): detection of specific instances versus the detection of broad categories. The first type aims to detect instances of a particular object (such as Donald Trump’s face, the Eiffel Tower, or a neighbor’s dog), essentially… | concept_edges[3] |
| weight | specific instances —distinct-from→ broad categories — 0.95— “As illustrated in Fig. 2, object detection can be grouped into one of two types (Grauman and Leibe 2011; Zhang et al. 2013): detection of specific instances versus the detection of broad categories. The first type aims to detect instances of a particular object (such as Donald Trump’s face, the Eiffel Tower, or a neighbor’s dog), essentially… | concept_edges[3] |
| differs in (stated by the tagger, no quotation) | accuracy —distinct-from→ efficiency — whether the constraint comes from the vision problem or from the deployment device | concept_edges[4] |
| relation | accuracy —distinct-from→ efficiency — the two competing goals - localising and recognising well enough to tell categories apart, against running in real time within a device's memory— “The ideal of generic object detection is to develop a generalpurpose algorithm that achieves two competing goals of high quality/accuracy and high efficiency (Fig. 5). As illustrated in Fig. 6, high quality detection must accurately localize and recognize objects in images or video frames, such that the large variety of object categories in… | concept_edges[4] |
| weight | accuracy —distinct-from→ efficiency — 0.95— “The ideal of generic object detection is to develop a generalpurpose algorithm that achieves two competing goals of high quality/accuracy and high efficiency (Fig. 5). As illustrated in Fig. 6, high quality detection must accurately localize and recognize objects in images or video frames, such that the large variety of object categories in… | concept_edges[4] |
| outcome | control — held— “Since the introduction of MS COCO, more attention has been placed on the accuracy of the bounding box location. Instead of using a fixed IOU threshold, MS COCO introduces a few metrics (summarized in Table 5) for characterizing the performance of an object detector. For instance, in contrast to the traditional mAP computed at a single IoU of… | controls[0] |
| outcome | control — held | controls[1] |
| outcome | control — held | controls[2] |
| outcome | control — held | controls[3] |
| outcome (stated by the tagger, no quotation) | control — absent-declared-unreported Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | controls[4] |
| limitation | whether a detection is correct depends on a predefined overlap threshold rather than on any property of the image Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | limitations[0] |
| limitation | mAP means different things under PASCAL VOC and MS COCO, so numbers are not comparable across that boundary Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | limitations[1] |
| limitation | the number of object categories in existing benchmark datasets is much smaller than can be recognized by humans - 20, 200 and 91 against a real-world order of 10^4-10^5 Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | limitations[2] |
| limitation | the challenge winner is decided by the number of object categories won, with mAP used as a team measure justified by the two rankings having always agreed Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | limitations[3] |
| limitation | high accuracy and high efficiency are competing goals, so any single reported figure is a position on a trade-off rather than a level of performance Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | limitations[4] |
finding · quantity · variable · definition · key-term — grounded in the full text| finding | Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | content_tags.findings[0] |
| finding | Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | content_tags.findings[1] |
| finding | Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | content_tags.findings[2] |
| finding | Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | content_tags.findings[3] |
| finding | Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | content_tags.findings[4] |
| finding | Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | content_tags.findings[5] |
| finding | Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | content_tags.findings[6] |
| quantity | of the objects in MS COCO are small and 24% are large - the reason size-stratified metrics exist Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | content_tags.quantities[0] |
| quantity | more than this many research contributions are included in the survey Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | content_tags.quantities[1] |
| variable | object categories Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | content_tags.variables[0] |
| variable | IOU Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | content_tags.variables[1] |
| definition | True Positive Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | content_tags.definitions[0] |
| definition | mAP Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | content_tags.definitions[1] |
| concept | IOU — not defined in this paper's own words; intersection over union of predicted and ground-truth boxes - the quantity a chosen threshold is applied to in order to decide whether a detection is correct at all Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | concepts[0] |
| concept | True Positive — not defined in this paper's own words; a detection whose category matches and whose overlap clears a predefined threshold - a verdict that depends on a number nobody derived Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | concepts[1] |
| concept | mAP — not defined in this paper's own words; mean average precision over categories - the field's headline number, and one whose definition differs between PASCAL VOC and MS COCO Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | concepts[2] |
| concept | MS COCO — not defined in this paper's own words; the benchmark that replaced the single-threshold metric with an average over ten thresholds, and added size-stratified metrics because 41% of its objects are small Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | concepts[3] |
| concept | object categories — not defined in this paper's own words; the class list - 20 in PASCAL VOC, 200 in ILSVRC, 91 in MS COCO, and the survey says all of them are much smaller than what humans recognise Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | concepts[4] |
| concept | appearance — not defined in this paper's own words; how an object looks - which varies astonishingly within a class and can be near-identical across classes, so it does not determine the label in either direction Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | concepts[5] |
| concept | object class — not defined in this paper's own words; the annotated label; the survey's own figure shows one class spanning wildly different images and four near-identical images spanning four classes Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | concepts[6] |
| concept | intraclass variations — not defined in this paper's own words; the differences among instances that share a label - intrinsic factors and imaging conditions, and the first of the two named accuracy challenges Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | concepts[7] |
| concept | accuracy — not defined in this paper's own words; one of the two competing goals; high quality means localising and recognising well enough that categories can be distinguished and instances of one category still recognised Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | concepts[8] |
| concept | efficiency — not defined in this paper's own words; the other competing goal - real time with acceptable memory and storage, which is a constraint from the deployment device rather than from the vision problem Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | concepts[9] |
| concept | specific instances — not defined in this paper's own words; detecting one particular object - the survey calls it essentially a matching problem, and separates it from the generic task Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | concepts[10] |
| concept | broad categories — not defined in this paper's own words; the generic detection problem, where the categories are given in advance by a dataset Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | concepts[11] |
| mAP | Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | content_tags.statistics[0] |
| IOU | Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | content_tags.statistics[1] |
When we have changed our mind. Some entries in the table begin with a word in capitals and the reason for it. EVIDENCE CORRECTED — the finding still stands, but we had quoted the wrong sentence from the paper, and the quotation has been replaced with the right one. RETRACTED — the entry itself was wrong; it stays on the page, marked, with the reason, rather than being deleted. WITHDRAWN AND REPLACED — the entry was wrong, and a corrected version of it appears in the entries below. Each mark carries its own reason, in brackets, because the reasons differ from one entry to the next. An entry we have WITHDRAWN is removed from this page and kept in the record, with its reason. A line reading Audit and a date is a later re-reading of the entry against its whole sentence: DOUBTFUL means it was kept, and the reason says what is loose about it. Quotation marks inside the tagger's sentence. Where a tag is the tagger's own sentence, a phrase in quotation marks inside it is borrowed from the paper — the tagger's wording around it is not. The paper's own sentence is the evidence quote for that row; on this page it is withheld with the article, and the article is at its publisher. Nothing inside quotation marks anywhere on this page is ours: if it is quoted, it is the paper's. Whose words a concept carries. A concept entry records a term the paper itself uses, so the term, the sense given for it, and its evidence are the paper's own words, quoted where the paper fixes them. This is the opposite of a sense entry in the tag table, which is the tagger's sentence about how the paper uses a term. Where a concept entry could not be grounded in the paper's words, the entry says so on its own row.
Tag layer © the Meta-Tagging Project, CC BY 4.0. The paper text is © its authors/publisher
and is linked, not reproduced. Excerpts shown are short quotations for identification and verification.
· back to index
· concepts · authors
· עברית
© 2026 Shir Sivroni · ORCID 0009-0000-2597-2824