Semi-supervised Learning for Dense Object Detection in Retail Scenes

Abstract

Retail scenes usually contain densely packed high number of objects in each image. Standard object detection techniques use fully supervised training methodology. This is highly costly as annotating a large dense retail object detection dataset involves an order of magnitude more effort compared to standard datasets. Hence, we propose semisupervised learning to effectively use the large amount of unlabeled data available in the retail domain. We adapt a popular self supervised method called noisy student initially proposed for object classification to the task of dense object detection. We show that using unlabeled data with the noisy student training methodology, we can improve the state of the art on precise detection of objects in densely packed retail scenes. We also show that performance of the model increases as you increase the amount of unlabeled data.

Penalizing Proposals using Classifiers for Semi-Supervised Object Detection

“…Self-training methods [27], [28], [29], [7] are another class of algorithms using pseudo labels generated by a soft teacher to train on a large number of unlabeled samples. There are many studies involving the generation of pseudo labels in the object detection literature [7], [8], [30], [31]. Some of these algorithms involve filtration of the pseudo labels based on some threshold [7], [8], to discard the labels that may not be useful.…”