Object detection in densely packed scenes is a new area where standard objectdetectors fail to train well. Dense object detectors like RetinaNet trained onlarge and dense datasets show great performance. We train a standard objectdetector on a small, normally packed dataset with data augmentation techniques.This dataset is 265 times smaller than the standard dataset, in terms of numberof 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 SKUproduct detection by providing full annotations for multiple public datasets.It can be accessed athttps://github.com/ParallelDots/generic-sku-detection-benchmark. We hope thatthis benchmark helps in building robust detectors that perform reliably acrossdifferent settings in the wild.