Benchmark for Generic Product Detection: A Low Data Baseline for Dense Object Detection
This provides a low-data baseline and benchmark for product detection in retail settings, but is incremental as it applies existing methods to new data.
The paper tackles the problem of object detection in densely packed scenes by training a standard object detector on a dataset 265 times smaller than standard datasets, achieving satisfactory results with mAP=0.56 at IoU 0.5, and creates a benchmark for generic SKU product detection.
Object detection in densely packed scenes is a new area where standard object detectors fail to train well. Dense object detectors like RetinaNet trained on large and dense datasets show great performance. We train a standard object detector on a small, normally packed dataset with data augmentation techniques. This dataset is 265 times smaller than the standard dataset, in terms of number of annotations. This low data baseline achieves satisfactory results (mAP=0.56) at standard IoU of 0.5. We also create a varied benchmark for generic SKU product detection by providing full annotations for multiple public datasets. It can be accessed at https://github.com/ParallelDots/generic-sku-detection-benchmark. We hope that this benchmark helps in building robust detectors that perform reliably across different settings in the wild.