SkillAgentSearch skills...

HeIvIndex

Content based image search (with hamming embedding)

Install / Use

/learn @yang2640/HeIvIndex
About this skill

Quality Score

0/100

Supported Platforms

Universal

README

Author: Yang Zhou

Contact: zhouyang2640@gmail.com

Build: Go to the hamming dir, type make, the executables will be placed to the HeIvIndex dir.

plug in the data, follow the pipeline:

(1) put all ukbench images under data/Images

(2) put all extracted feature files under data/Images, the format of the feature file follow CSV (space separated)

(3) sh sample.sh, construct the training data for kmeans

(4) python genProjectMatrix.py, construct the projection matrix for hamming embedding

(5) ./kmeans, build the codebook

(6) ./he_medians, construct the median mat for hamming embedding

(7) ./he_quantize, do a batch quantization for all images, feature paths are listed in data/featlist

(8) ./he_query > query.txt, do a batch query for all images, feature paths are listed in data/featlist

(9) python computeNSScore.py, to compute the NS score

Note:

(1) A lot of files are pre-loaded, I have not checked the read error, e.g., the file doesnot exist

(2) All the generated files are put in the data dir

Related Skills

View on GitHub
GitHub Stars8
CategoryContent
Updated4y ago
Forks1

Languages

Shell

Security Score

55/100

Audited on Mar 17, 2022

No findings