Man-made workpiece counting is a routine job for manufactory workers; however, this is an error-prone task. In this paper, we are interested in detecting and counting arbitrary workpieces in industrial manufacturing. Therefore, we construct a comprehensive and large-scale open-world public benchmark dataset for workpiece counting, called Workpiece Counting Dataset (WPCD), which includes 121,475 instances of workpieces from 351 different categories. We also propose a novel method for workpiece detection and counting, named Two-stage Workpiece Counting Network (TS-WPCNet). The first stage of the network is to develop a class-agnostic detector to localize each workpiece instance, followed by the second stage was to employ an Unsupervised Deep Clustering Strategy (UDCS) with the backbone network pre-trained in a Workpiece Convolutional AutoEncoder (WCAE) for decision boundary prediction, achieving workpiece clustering under unknown K values. Finally, our experiments show that the proposed method outperforms current mainstream methods, greatly enhancing the efficiency of factory operations.
Two-stage Workpiece Counting Network. The network aims to detect and count arbitrary workpieces in industrial manufacturing. In the first stage, a class-agnostic detector is designed by incorporating the CALM module to accurately locate each instance of the workpieces. In the second stage, an unsupervised deep clustering strategy is introduced to achieve workpiece clustering under unknown K conditions, with the backbone network pretrained on WCAE. Finally, the statistical results of each workpiece class and overall count are obtained.