1D Barcode Detection: Novel Benchmark Datasets and Comprehensive Comparison of Deep Convolutional Neural Network Approaches

Abstract

Recent advancement in Deep Learning-based Convolutional Neural Networks (D-CNNs) has led research to improve the efficiency and performance of barcode recognition in Supply Chain Management (SCM). D-CNNs required real-world images embedded with ground truth data, which is often not readily available in the case of SCM barcode recognition. This study introduces two invented barcode datasets: InventBar and ParcelBar. The datasets contain labeled barcode images with 527 consumer goods and 844 post boxes in the indoor environment. To explore the influential capability of the datasets that affect recognition process, five existing D-CNN algorithms were applied and compared over a set of recently available barcode datasets. To confirm the model’s performance and accuracy, runtime and Mean Average Precision (mAP) were examined based on different IoU thresholds and image transformation settings. The results show that YOLO v5 works best for the ParcelBar in terms of speed and accuracy. The situation is different for the InventBar since Faster R-CNN could allow the model to learn faster with a small drop in accuracy. It is proven that the proposed datasets can be practically utilized for the mainstream D-CNN frameworks. Both are available for developing barcode recognition models and positively affect comparative studies.

An Encoder–Decoder Architecture within a Classical Signal-Processing Framework for Real-Time Barcode Segmentation

“…One-dimensional barcode detection: novel benchmark datasets and comprehensive comparison of deep convolutional neural network approaches [24] Following another work by the same group [20], where they systematically analysed the literature on neural networks applied to the problem under discussion, the authors created two training datasets: one containing consumer goods codes and the other containing postal labels. Both were created based on real, uncontrolled environments.…”

“…Moreover, it was the only one performing under the 33 ms time limit. Neural network-based methods were represented by the hand-crafted method by Zharkov and Zagaynov [17] and by our measurement of inference times of YOLO v5 small, as it was selected by Kamnardsiri et al [24] as a good representative of the state-of-the-art methods in terms of both accuracy and speed. Previous methods that manage to get below the 33 ms threshold are able to accomplish this because they target desktop GPUs, which this paper does not consider, as they are not a realistic platform for wearable AR devices.…”

“…Table 4. Accuracy metric results for our methods, those in Table 3, and those considered state-of-theart methods in [24] on two barcode datasets.…”

Refined Plane Segmentation for Cuboid-Shaped Objects by Leveraging Edge Detection

“…Kamnardsiri et al [Kam+22] perform a case study analyzing five different Artificial Neural Network (ANN) architectures. They present two new datasets, InventBar and ParcelBar with 527 and 844 images, respectively.…”

“…Especially barcode detection has been studied thoroughly and numerous datasets are publicly available. While Kamnardsiri et al [Kam+22] performed an analysis for a selection of algorithms, it would be interesting to analyze more diverse scenarios similar to Brylka et al [BSB20]. Other fields lack the availability of diverse datasets and the effective use of synthetic data can be investigated.…”