Mining has traditionally been a volatile industry that experiences significant fluctuations in profitability. Changes in demand globally can have dramatic effects on the bottom line of many mining companies. Stimulus spending in response to the Covid-19 pandemic has caused a spike in prices for metals and other raw materials, but this only serves to underline the unpredictability inherent to the industry. In addition many profitable mines are at the point of reaching maturity, leading to lower quality ores and longer transportation distances from deep mines.
Faced with an uncertain future industry leaders are turning to AI to provide some stability. Computer vision based applications can provide an edge for mining companies by increasing efficiency of mining operations and reducing costs. This blog will focus on three emerging use cases for AI in this sector. We will also show how these promising technologies can be better supported with high quality image and video annotation services.
Automatic ore and compound recognition
In order to find new deposits of valuable raw materials beneath the earth's surface experts must examine rock samples and core deposits. By analysing these pieces of evidence it is possible to recognize specific ore samples or other compounds that might indicate the presence of target commodities. Of course this kind of work can be painstaking, requiring thousands of sample images to be analysed. It also demands extensive expertise that can be costly to train and recruit.
AI applications have shown the capacity to streamline this important process. Exposing machine learning models to annotated images of various samples allows them to learn how to identify relevant compounds. This can significantly accelerate surveying for mining operations, allowing experts to focus on the most important samples, or ones requiring additional investigation.
This helpful use case is dependent on access to annotated image training data. However, due to the specialised knowledge needed to interpret these compounds it can be difficult to assemble a comprehensive dataset. Professional annotation services have the experience necessary to assemble bespoke images, backed up by specialist verification experts.
Mining operations often cover enormous areas. The world's largest mine, Bingham Canyon Mine in Utah, is around 4km wide and 1.2km deep. These huge sites need to be monitored for a number of reasons, including checking the stability of waste piles, ensuring the integrity of pipelines, and guarding against environmental damage.
Drones are now being considered as an effective means of overseeing mine operations and keeping track of a complex, changing environment. AI systems mean that these drones can be automated, functioning as an eye in the sky that gives managers and workers early warning of emerging issues.
Quality, annotated video imagery can give developers of this technology an edge. Professional annotation providers, like Keymakr, can make use of proprietary annotation tools that speed up the video labeling process. The right annotation platform can ensure that video training data keeps pace with research.
Autonomous mining machines
The final step for automation in the mining sector is for computer vision systems to take on some of the burden of mining itself. As mines are cut deeper and deeper the environment becomes increasingly dangerous for humans. AI powered robots and excavation vehicles can go to the most hazardous places and operate around the clock. This has the potential to increase safety for miners and increase productivity for mining companies.
To ensure the functionality of these autonomous vehicles it is important that they be trained with data that is as precise as possible. Annotation service providers are often best placed to ensure data accuracy because they employ rigorous quality control processes.
At Keymakr this includes three levels of human verification and an additional automated check. Contact a team member to book your personalized demo today.