Notice
Recent Posts
Recent Comments
Link
일 | 월 | 화 | 수 | 목 | 금 | 토 |
---|---|---|---|---|---|---|
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 |
Tags
- Average Precision
- fine tune AlexNet
- mean Average Precision
- Region proposal Network
- Object Detection metric
- Object Detection
- object queries
- Bounding box regressor
- multi task loss
- Map
- Hungarian algorithm
- RPN
- Non maximum suppression
- pytorch
- herbwood
- RoI pooling
- IOU
- Anchor box
- Multi-task loss
- YOLO
- Faster R-CNN
- Darknet
- Linear SVM
- AP
- R-CNN
- hard negative mining
- BiFPN
- Detection Transformer
- detr
- Fast R-CNN
Archives
- Today
- Total
목록BiFPN (1)
약초의 숲으로 놀러오세요

이번에는 CVPR 2020년에 발표된 EfficientDet 논문(EfficientDet: Scalable and Efficient Object Detection)을 읽고 리뷰해보도록 하겠습니다. 당시 object detection 모델은 모두 속도와 정확도 사이의 트레이드오프(trade-off)가 존재했습니다. EfficientDet은 효과적으로 모델의 구조를 변형함으로써 속도와 정확도라는 두 마리 토끼를 모두 잡았으며, resource에 맞게 모델의 scale을 자유롭게 조정할 수 있는 높은 scalability를 보였습니다. Research gap 정확도가 높은 object detection 모델은 model size가 크며 computational cost가 높아 real-world applica..
Computer Vision/Object Detection
2022. 12. 12. 17:55