Friday 31 January 2025
In recent years, artificial intelligence (AI) has become an essential part of our daily lives. From virtual assistants like Siri and Alexa to self-driving cars and medical diagnosis systems, AI has made significant strides in transforming various industries. However, one area where AI still lags behind is image recognition, particularly when it comes to detecting anomalies.
To address this challenge, researchers have turned to quantum computing, a relatively new field that uses the principles of quantum mechanics to perform calculations exponentially faster than classical computers. In a recent study, scientists from Japan’s Toppan Holdings used quantum computing to develop a novel method for detecting anomalies in images.
The team focused on detecting internal vine cracks in apples, which can be difficult to spot even by human inspectors. They created a dataset of binary images featuring normal and anomalous apples, including those with browning and vine cracks. The researchers then applied principal component analysis (PCA) to extract features from the images.
Next, they generated quantum kernels using these features and embedded them into support vector machines (SVMs), a popular machine learning algorithm. This allowed the team to create a learning model that could predict whether an apple was normal or anomalous based on its image.
The results were impressive. When compared to classical kernel SVMs, the quantum kernel SVMs achieved higher F1 scores and better discriminative power. The study also found that as the feature value increased, the increase in F1 score for the quantum kernel was smaller than that of the classical kernel. This suggests that quantum kernels may provide an advantage in certain situations.
The researchers used a quantum simulator to test their model, which generated similar results to those obtained using a real quantum computer. However, when they attempted to run the model on the actual quantum computer, errors accumulated due to noise and circuit depth limitations.
Despite these challenges, the study demonstrated the potential of quantum computing in image recognition applications. The team’s findings could have significant implications for industries that rely heavily on anomaly detection, such as food inspection, medical diagnosis, and quality control.
In the future, researchers will need to address the issue of noise and error accumulation in quantum computers to unlock their full potential. However, the promise of quantum computing in image recognition is undeniable, and this study has taken an important step towards realizing that potential.
Cite this article: “Quantum Computing Enhances Anomaly Detection in Image Recognition”, The Science Archive, 2025.
Artificial Intelligence, Quantum Computing, Image Recognition, Anomaly Detection, Apples, Internal Vine Cracks, Principal Component Analysis, Support Vector Machines, Machine Learning, Noise Accumulation.
Reference: Takao Tomono, Kazuya Tsujimura, “Quantum kernel learning Model constructed with small data” (2024).
