PFDet: 2nd Place Solution to Open Images Challenge 2018 Object Detection Track

This work addresses the problem of efficient and scalable object detection for large datasets, but it is incremental as it builds on existing methods to achieve competitive results in a specific competition.

The team tackled large-scale object detection on the Open Images dataset by developing a system that trains with 512 GPUs, handles sparsely verified classes and class imbalance, achieving 2nd place in the 2018 Kaggle challenge.