AI Revolutionizes Science: Counting with Precision and Speed

Artificial intelligence (AI) is transforming the slow, labor-intensive, and costly process of counting tiny objects in scientific research.

Introduction: Artificial intelligence (AI) has become a game-changer in various industries, and science is no exception. The use of AI in scientific research is revolutionizing the way we count and identify objects, bringing greater accuracy, speed, and scale to tasks that were once onerous for scientists. By leveraging machine learning algorithms, researchers are now able to overcome the limitations of human perception and automate the counting process. In this article, we explore two fascinating examples where AI is being used to count hairs on cotton leaves and microscopic cells of harmful algae, showcasing the potential of this technology in the field of science. Counting hairs on cotton leaves: Traditionally, experts in commercial breeding programs manually scored the hairiness of cotton leaves, impacting insect resistance, fiber yield, and value. However, this process was subjective and time-consuming. By developing AI models, scientists have achieved a 95% accuracy rate in generating hairiness scores comparable to human experts. Dr. Moshiur Farazi, an expert in computer vision, explains that their latest model, HairNet2, goes beyond automating current methods. It estimates the area of the leaf covered by hairs by locating every single hair, a task that is difficult for humans but achievable for AI. HairNet2 was trained using annotated images, resulting in a more robust, reliable, and accurate scoring system. The new models are now being deployed on a web interface for breeders to test during the next cotton season. Counting microscopic algae cells: Identifying harmful algal blooms, which can be toxic to humans and animals, is a time-consuming and labor-intensive process. Scientists traditionally rely on microscopic examination and manual counting to estimate the number of harmful algal cells in a liquid sample. To address this challenge, researchers are training machine learning models to automatically detect harmful algae in images. By systematically photographing and annotating algae strain samples, an annotated dataset has been created for 15 different strains. Early testing shows that these AI models can successfully detect the target strains with high accuracy. The use of AI for faster and more accurate detection of toxic algae could have significant economic, environmental, and social impacts, providing early warning signals for water managers and safeguarding the health of coastal communities, consumers, and fisheries and aquaculture businesses. Conclusion: The integration of AI into scientific research is transforming the way we count and identify objects. From counting hairs on cotton leaves to identifying harmful algae cells, AI is bringing greater precision, speed, and scalability to tasks that were once slow, labor-intensive, and costly. The success of AI implementation in object detection depends on asking the right questions and preparing the data appropriately. As organizations explore the potential of AI in their processes, the barrier to entry has become lower, making it more accessible to businesses. The use of AI in science is not only revolutionizing the way we work but also opening up new possibilities for discovery and innovation.