What are the steps to convert 1000 of images into vectors for FAISS?
This is task-dependent, i.e. it really depends on what you want to do with you images. It is somehow how of scope for Faiss, but out of curiosity, what is the problem you want to address?
On 12 May 2017, at 01:36, zoe-cheung <[email protected]notifications@github.com> wrote:
What are the steps to convert 1000 of images into vectors for FAISS?
—
You are receiving this because you are subscribed to this thread.
Reply to this email directly, view it on GitHubhttps://github.com/facebookresearch/faiss/issues/114, or mute the threadhttps://github.com/notifications/unsubscribe-auth/AY5yTQMKPmGXXhJiXfyg_3c0j-_-DTV6ks5r45QJgaJpZM4NYqGz.
FAISS needs vector dataset. I have image files. I just want to convert the images into vectors for FAISS. I would like to match similar images.
Converting an image into a vector is irrelevant to Faiss. You need to study CNN or SIFT to extract features(vector) from images. Then you can use Faiss to index these vectors for searching or matching.
Most helpful comment
Converting an image into a vector is irrelevant to Faiss. You need to study CNN or SIFT to extract features(vector) from images. Then you can use Faiss to index these vectors for searching or matching.