LLA: Loss-aware Label Assignment for Dense Pedestrian Detection
This work addresses label assignment specifically for dense pedestrian detection in crowd scenarios, offering a simple yet effective method that is incremental over existing general object detection techniques.
The paper tackles the problem of label assignment in dense pedestrian detection by proposing Loss-aware Label Assignment (LLA), which uses joint loss as an assigning indicator to select positive anchors, resulting in improvements of 9.53% and 5.47% in MR on RetinaNet and FCOS detectors, respectively.
Label assignment has been widely studied in general object detection because of its great impact on detectors' performance. However, none of these works focus on label assignment in dense pedestrian detection. In this paper, we propose a simple yet effective assigning strategy called Loss-aware Label Assignment (LLA) to boost the performance of pedestrian detectors in crowd scenarios. LLA first calculates classification (cls) and regression (reg) losses between each anchor and ground-truth (GT) pair. A joint loss is then defined as the weighted summation of cls and reg losses as the assigning indicator. Finally, anchors with top K minimum joint losses for a certain GT box are assigned as its positive anchors. Anchors that are not assigned to any GT box are considered negative. Loss-aware label assignment is based on an observation that anchors with lower joint loss usually contain richer semantic information and thus can better represent their corresponding GT boxes. Experiments on CrowdHuman and CityPersons show that such a simple label assigning strategy can boost MR by 9.53% and 5.47% on two famous one-stage detectors - RetinaNet and FCOS, respectively, demonstrating the effectiveness of LLA.