Improve pedestrian detection by using general object detector with Cascade RCNN

Improve pedestrian detection by using general object detector with Cascade RCNN

Pedestrian Detection: Domain Generalization, CNNs, Transformers and Beyond
arXiv paper abstract https://arxiv.org/abs/2201.03176v1
arXiv PDF paper https://arxiv.org/pdf/2201.03176v1.pdf
GitHub https://github.com/hasanirtiza/Pedestron

… current pedestrian detectors poorly handle even small domain shifts in cross-dataset evaluation.

… attribute the limited generalization to two main factors, … biasness present in the design choices (e.g anchor settings) … Most modern pedestrian detectors are tailored towards target dataset

… Consequently, a general object detector performs better in cross-dataset evaluation compared with state of the art pedestrian detectors

… benchmarks curated by crawling the web (which contain diverse and dense scenarios), are an efficient source of pre-training for providing a more robust representation.

… propose a progressive fine-tuning strategy which improves generalization. …

Stay up to date. Subscribe to my posts https://morrislee1234.wixsite.com/website/contact
Web site with my other posts by category https://morrislee1234.wixsite.com/website

LinkedIn https://www.linkedin.com/in/morris-lee-47877b7b

Photo by Vlad B on Unsplash