티스토리 뷰

[업데이트 2018.09.06 13:42]


1. 베이스 논문

[논문 요약17] Perceptual Generative Adversarial Networks for Small Object Detection

(http://arclab.tistory.com/172)


The hand-crafted features achieve great success in pedestrian detection. For example, Doll´ar et al. proposed Integral Channel Features(ICF) [8] and Aggregated Channel Features (ACF) [7], which are among the most popular hand-crafted features for constructing pedestrian detectors.


For the Caltech benchmark [9], we utilize the ACF pedestrian detector [7] trained on the Caltech training set for object proposals generation.


2. 레퍼런스 논문

[7] P. Doll´ar, R. Appel, S. Belongie, and P. Perona. Fast feature pyramids for object detection. TPAMI, 36(8):1532–1545, 2014. 2, 5

(https://vision.cornell.edu/se3/wp-content/uploads/2014/09/DollarPAMI14pyramids_0.pdf)


3. 주요 내용 요약

3.1 레퍼런스 논문 주요 내용

종래의 최신 기술의 경우 computational bottleneck이 존재하였는데, 잘 샘플링된 image pyramid의 모든 scale에 대해 feature를 계산하기 때문이었습니다. 본 논문에서는 모든 scale에 대해 다루지 않고, interval을 두어 샘플링된 image pyramid에 대해 근사화를 하여 성능 향상 시키는 방법을 제안하고 있습니다.



3.2 베이스 논문에서 인용한 내용

- Deep learning 이전에는 보행자 인식 분야에 있어서 hand-crafted된 feature를 통하여 좋은 성과를 낸 경우가 있었습니다.

  (proposed Integral Channel Features (ICF) [8] 와 Aggregated Channel Features (ACF) [7])

 

- Caltech benchmark(보행자 데이터셋)를 위해 Caltech training set으로 학습된 ACF pedestrian detector [7]를 사용하였습니다.


댓글
공지사항
최근에 올라온 글
최근에 달린 댓글
Total
Today
Yesterday
링크
«   2024/05   »
1 2 3 4
5 6 7 8 9 10 11
12 13 14 15 16 17 18
19 20 21 22 23 24 25
26 27 28 29 30 31
글 보관함