Deep Learning for Retail Product Recognition: Challenges and Techniques

Table 7

Product detection benchmark results on the test set of the D2S dataset.

Approaches	mAP (%)	(%)	(%)
Mask R-CNN [55]	78.3	89.8	84.9
FCIS [93]	68.3	88.5	80.9
Faster R-CNN [50]	78.0	90.3	84.8
RetinaNet [159]	80.1	89.6	84.5