search

albanie / collaborative-experts

Video embeddings for retrieval with natural language queries
https://www.robots.ox.ac.uk/~vgg/research/collaborative-experts/
Apache License 2.0
336 stars 55 forks source link

Query search #32

Closed Tortoise17 closed 3 years ago

Tortoise17 commented 3 years ago

You have mentioned the query search of video search and similarity score in the paper. Is it possible to use the command line with the available code here in github? because in the ReadMe there is no info.

Tortoise17 commented 3 years ago

Query search means based on the score of the model evaluation of video how you have mentioned in paper.

Tortoise17 commented 3 years ago

@ioanacroi can you suggest anything on it?

ioanacroi commented 3 years ago

Hi,

Can you give more details (e.g where in the paper etc)? It's not clear for me what you mean by "query search of video search".

Thank you, Ioana

Tortoise17 commented 3 years ago

@ioanacroi page 10 in this paper. https://arxiv.org/pdf/1907.13487.pdf

Tortoise17 commented 3 years ago

Normally with current visualization is also super cool. I meant is there is any commandline additionally or tool availablw which is providing freedom to enter the text in query and getting out the videos with scores as api or cli.

ioanacroi commented 3 years ago

Our code outputs the score between each query and each video in form of a similarity matrix. In the fig.3 of the paper you pointed out, you can see the top 3 video retrieved by each query based on the similarity score (which were extracted from the similarity matrix). The current code outputs the similarity matrix for all the queries and all the video. We do not have an API where you can enter the text query and get the score for the videos or the video names.

Cheers, Ioana

Tortoise17 commented 3 years ago

@ioanacroi But overall, this study is a great study. I must admit, that the code works excellent. The research is very good.I liked the way it works. Great piece of work.

Tortoise17 commented 3 years ago

r