These are chat archives for beniz/deepdetect

25th
May 2016
deter3
@deter3
May 25 2016 02:27
This message was deleted
deter3
@deter3
May 25 2016 02:41
@beniz I am playing with image similarity search demo now , I am wondering wether I can limit the image names to search for ? say I want to search in image_food1 , image_food2 ,image_food3 ......... for image_rice picture ? I want to increase the search effieincy by narrowing down the index to search .
Emmanuel Benazera
@beniz
May 25 2016 16:13
I'm not sure what you mean by efficiency, whether it's speed or accuracy
deter3
@deter3
May 25 2016 16:56
@beniz actually both , mainly accuracy .
Emmanuel Benazera
@beniz
May 25 2016 17:28
then create many indexes instead of one. but the real way of getting great accuracy is to finetune your neural accordingly, in order to get short and accurate binary codes
deter3
@deter3
May 25 2016 17:51
@beniz thanks for the reply . That is true , fine-tune my neural , that's my next step . Again , I want to thank you for your building up this software , which make it much easier to get into deep learning . I read a lot and watch a lot of videos , only you are providing detailed hands on & entry level material and plug & play software for deep learning . One more suggestion , if you can record videos (3-5 mins ) , together with your tutorials , will help a lot people . i guess I will need to talk to you on the possibility of professional consultant very soon if possible .
Emmanuel Benazera
@beniz
May 25 2016 19:33
Thanks for the kind words. I'd let community provide the videos, don't hesitate if you are interested.
I mean interested in providing video tutorials
Do not hesitate to propose new tutorials as well, documentation is an effort in itself