Open With-the-sun opened 2 years ago
hi The Jaccard distance was uesd in minhash for your cases, so, does it means that feature vectors extracted from the image cannot replace datasets set1,set2 and set3 (in "datasketch-master/examples/lsh_examples.py")?
I am a beginner and want to use LSH to quickly query images or other types of data. Ask if only text data is included in the project
WeightMinHash can be used for images, but I think feature engineering is required to prepare the input vectors.
hi The Jaccard distance was uesd in minhash for your cases, so, does it means that feature vectors extracted from the image cannot replace datasets set1,set2 and set3 (in "datasketch-master/examples/lsh_examples.py")?
For MinHash in this package, it is designed to work with textual data. So it's not recommended to use image data here.
You can try WeightedMinHash instead.
Any skeleton code, perhaps with openCV SIFT descriptors?
I am a beginner and want to use LSH to quickly query images or other types of data. Ask if only text data is included in the project