Simple Training Strategies and Model Scaling for Object Detection
By Xianzhi Du, Barret Zoph, Wei-Chih Hung, Tsung-Yi Lin. 2021
52 citations recorded by the source corpus.
The speed-accuracy Pareto curve of object detection systems have advanced through a combination of better model architectures, training and inference methods. In this paper, we methodically evaluate a variety of these techniques to understand where most of the improvements in modern detection systems come from. We benchmark these improvements on the vanilla ResNet-FPN backbone with RetinaNet and RCNN detectors. The vanilla detectors are improved by 7.7% in accuracy while being 30% faster in speed. We further provide simple scaling strategies to generate family of models that form two Pareto curves, named RetinaNet-RS and Cascade RCNN-RS. These simple rescaled detectors explore the speed-accuracy trade-off between the one-stage RetinaNet detectors and two-stage RCNN detectors. Our largest Cascade RCNN-RS models achieve 52.9% AP with a ResNet152-FPN backbone and 53.6% with a SpineNet143L backbone. Finally, we show the ResNet architecture, with three minor architectural changes, outperforms EfficientNet as the backbone for object detection and instance segmentation systems.
Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition, Optimization & Training
How researchers see this paper
The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.
In Artificial Intelligence
- Rohit Girdhar
- Ishan Misra
- Alexander Kirillov
- Alexander Gerhard Schwing
- Hongpeng Liao
- Chien-Yao Wang
- Alexey Bochkovskiy
- Bowen Cheng
In Computer Vision and Pattern Recognition
- Luc Van Gool
- Jitendra Malik
- Dacheng Tao
- Alan L. Yuille
- Xiaogang Wang
- Ming-Hsuan Yang
- Andrew Zisserman
- Rohit Girdhar
In Optimization & Training
- Alexander Kirillov
- Alexey Bochkovskiy
- Hongpeng Liao
- Ishan Misra
- Alexander Gerhard Schwing
- Chien-Yao Wang
- Rohit Girdhar
- Bowen Cheng
Citations in the corpus
- Masked-attention Mask Transformer for Universal Image Segmentation (2021, methodology): “With ResNet [25] backbone, Mask2Former outperforms a strong Mask R-CNN [24] baseline using large-scale jittering (LSJ) augmentation [18, 23] while requiring 8 × fewer training iterations.”
- YOLOv7: Trainable Bag-of-Freebies Sets New State-of-the-Art for Real-Time Object Detectors (2022, background): “Model scaling [72, 60, 74, 73, 15, 16, 2, 51] is a way to scale up or down an already designed model and make it fit in different computing devices.”