开源Python高维数据最近邻快速检索库：Panns
cymt
9年前
Panns=Python Approximate Nearest Neighbor Search，基于索引优化技术，支持欧氏距离和余弦夹角作为相似度测度，支持并行查询。
特性
- Pure python implementation.
- Optimized for large and high-dimension dataset (e.g. > 500).
- Generate small index file with high query accuracy.
- Support both Euclidean and cosine distance metrics.
- Support parallel building of indices.
- Small memory usage and index can be shared among processes.
- Support raw, csv, numpy and HDF5 datasets.