FACE DATASET

[sharad1999]

[sharad1999]

Can you help me how to train the face dataset

[Kyfafyd]

How your face dataset is formulated?

[sharad1999]

It is the lfw dataset

[sharad1999]

How your face dataset is formulated?

It has captions.json

[sharad1999]

How your face dataset is formulated?

It also has captions.pickle but no train.pickle

[Kyfafyd]

Can you show me some more details of your current status? e.g. the dataset url. code .etc Sorry for replying so late, I am on several urgent projects...

[sharad1999]

Can you show me some more details of your current status? e.g. the dataset url. code .etc Sorry for replying so late, I am on several urgent projects...

yes sir so i want to implement the LFW dataset it has 400 images & has a captions.json file which is like this

[ { "img_id": 1, "image": "Katerina_Smrzova/Katerina_Smrzova_0001.bmp", "descriptions": [ { "desc_id": 582, "text": "A young woman with long light brown hair cut in layers and parted in the middle, a small nose and a nice open smile." }, { "desc_id": 1326, "text": "A young girl posing for the camera" } ] }, { "img_id": 2, "image": "John_Rowland/John_Rowland_0001.bmp", "descriptions": [ { "desc_id": 178, "text": "Middleaged man - very traditional square jawed american features looking pensive or waiting for a result of somesort. looks like a politician" }, { "desc_id": 851, "text": "brown/greyish hair, brown eyes" } ] },

This json file is a cleaned file which i got from a repo. & the images & texts can be found here

http://vis-www.cs.umass.edu/lfw/

I have the captions.pickle file from https://github.com/akanimax/T2F but for my training i need train.pickle & test.pickle as well

[sharad1999]

so sir bascially i want my data to be in the format of birds dataset

[sharad1999]

also sir i dont understand how does the model know that the particular text belongs to this image & for that particular text these are the embeddings of that image