A Measurement Method of Mineral Raw Material Intensity Based on Image Object Detection

In the metallurgical industry, online monitoring of belt conveyors' mineral raw material flow intensity helps avert severe accidents during production. There is no established method for detecting mineral raw material flow intensity in real time using computer-based technologies. So, we proposed two computer vision processing methods: traditional image processing algorithm approach and a CNN (Convolutional Neural Networks) MobileNetV2-SSD (Single Shot MultiBox Detector) detection method based on deep learning. We created a dataset containing 5731 images and experimented it on both methods, and also assessed it on different convolutional neural network models. In summary, the conventional image processing algorithm holds an accuracy of 81.68%, whereas the MobileNetV2- SSD model exhibited 99.65% accuracy. The MobileNetV2-SSD reflected significant advantages over other models in terms of accuracy and time consumption. So, we conclude that MobileNetV2-SSD mineral raw material flow intensity detection model can be widely used in metallurgical production to automate the production process and improve productivity.