MicroCloud Hologram Inc. (NASDAQ: HOLO) announced the development of hybrid quantum-classical three-dimensional object technology for multi-channel quantum convolutional neural networks, according to a company press release.

The technology combines quantum and classical computing approaches for 3D object detection tasks. The system uses classical computing for sensor data preprocessing and 3D point cloud construction, while quantum computing handles convolutional feature extraction processes.

The company's Multi-Channel Quantum Convolutional Neural Network (MC-QCNN) maps multi-channel 3D feature maps into quantum state space through quantum state entanglement mechanisms. The system uses parameterized quantum circuits to construct quantum convolution kernels defined by trainable quantum gate parameters.

MicroCloud Hologram implemented a knowledge distillation mechanism during model training, using a classical 3D object detection model as a teacher model and the hybrid quantum-classical detection model as a student model. The approach aims to achieve detection accuracy comparable to classical models while operating under quantum resource constraints.

The technology is designed for noisy intermediate-scale quantum devices rather than large-scale fault-tolerant quantum computers. The company states this makes the technology deployable in the near term while allowing for future expansion as quantum hardware improves.

MicroCloud Hologram indicated the technology could extend to other 3D computer vision tasks including point cloud semantic segmentation and multi-sensor fusion perception. The company plans to continue development for industrial implementation.

MicroCloud Hologram provides holographic technology services including LiDAR solutions and holographic imaging. The company reported cash reserves exceeding $390 million and plans to invest over $400 million in quantum computing research and development.