2020
DOI: 10.1007/978-3-030-68238-5_23
|View full text |Cite
|
Sign up to set email alerts
|

The 1st Tiny Object Detection Challenge: Methods and Results

Help me understand this report

Search citation statements

Order By: Relevance

Paper Sections

Select...
3
2

Citation Types

0
10
0

Year Published

2021
2021
2024
2024

Publication Types

Select...
5
2

Relationship

1
6

Authors

Journals

citations
Cited by 27 publications
(13 citation statements)
references
References 15 publications
0
10
0
Order By: Relevance
“…Despite the great success of the generic object detection methods trained on ground-to-ground images, a huge performance drop is observed when they are directly applied to images captured by UAVs [36], [171]. Examples of the performance drop of state-of-the-art detectors are illustrated in Fig.…”
Section: Challenges For Aerial Human Detectionmentioning
confidence: 99%
See 2 more Smart Citations
“…Despite the great success of the generic object detection methods trained on ground-to-ground images, a huge performance drop is observed when they are directly applied to images captured by UAVs [36], [171]. Examples of the performance drop of state-of-the-art detectors are illustrated in Fig.…”
Section: Challenges For Aerial Human Detectionmentioning
confidence: 99%
“…The Cascade R-CNN [17] drops the performance by 50% in the aerial VisDrone [187] dataset compared with the ground-based COCO and Pascal VOC datasets [36]. The Faster R-CNN [117] also drops the performance by 30% in the aerial TinyPersons [173] dataset compared with the ground-based COCO and Pascal VOC datasets [171].…”
Section: Challenges For Aerial Human Detectionmentioning
confidence: 99%
See 1 more Smart Citation
“…In a recent survey [42], the authors categorize numerous approaches to CNN-based object detection as one-stage [59] vs. two-stage [39] architectures, single-scale [59] vs. pyramid feature networks [38], and anchor-based [44] vs. anchor-free [19,6,80,18] tech- niques. These approaches focus on model refinement for a single detector, but their performance is often stifled by their model's capacity for realistic tasks [74].…”
Section: Introductionmentioning
confidence: 99%
“…As demonstrated by numerous engineering attempts and visual object detection contests [74], single detector approaches still do not perform well enough for real applications. It is our belief, however that detector ensemble approaches can push the boundaries of object detection performance to acceptable levels.…”
Section: Introductionmentioning
confidence: 99%