The dataset is the official algorithm library for the Tianmouc complementary vision sensor, providing tools for reading, processing, and visualizing sensor data from the first multi-pathway brain-inspired vision sensor. It contains codecs for Tianmouc data files (.tmdat), deep learning dataset readers, ISP tools, feature descriptions for sparse temporal and spatial differences, optical flow estimators, gray-scale reconstructors, and neural networks for reconstruction, optical flow, deblurring, and instance segmentation. The purpose is to support high-speed (10000 fps), high-dynamic-range (130 dB), high-sensitivity visual perception with three data modalities: frame-based RGB, sparse Temporal Difference, and sparse Spatial Differences, aimed at enabling complementary vision sensor applications in computer vision and robotics.