The story of how we did the neural network to remove the background, caught up with the quality of the market leader and highlighted the resulting BENZIN neural network in a separate product - https://benzin.io At the end of the article will discuss the results for the calendar year, work on our mistakes, and what are the prospects for this project. It will be useful for those who are planning or already creating their own SaaS.
History of creation
Last year, right before the quarantine, I was selling an online store project. And in order to differentiate ourselves from competitors, we offered our customers such a feature for free - a pipeline loading of photos from a camera directly into an accounting system. I understand that these photos are not really needed in 1C, but the fact is that in 1C there is a setting for external storage of files and standard unloading of goods in 1C Bitrix and the main CMS for online stores. Accordingly, this greatly speeds up the process of filling an online store with product photos, especially if there are a lot of products. We are talking about tens of thousands.
In comparison with the traditional method of uploading photos, our conveyor loading allows you to process from 80 to 120 items per day, 4-5 photos per item. If you think that filling 2.5 thousand products with original photos a month in an online store per month is slow, then the usual way: camera -> flash drive -> CMS will be several times slower, because most of the time will be lost on searching for a product in the catalog or comparing a photo of a product on a USB flash drive. Nobody pays special attention to this, since on the one hand this is not a problem for programmers and not for subject photographers. But it takes a lot of time.
For whom is this critical? Accelerating the photography process is not necessary for everyone. But imagine a shoe business, or a clothing business. Several new collections are released every year, and each product must be photographed and placed in the catalog. It is the fusion of the process of photographing and placing in the catalog (1C) that gives such a powerful acceleration effect. I hope you don’t need to explain the importance of original photographs for an online catalog or online store.
. 1 , . , , . , Canon, , Nikon. Nikon.
: , ? -, , . : ? , - . , . , . .
, "" - remove.bg, , . , , , . . , , - .
. : . - , . . , , , . remove bg. , , , .
, . 100% , , c. . 20 , / Gigabyte GeForce GTX 1070 Ti Gaming 8G. .
, . , Canon, Nikon. 1. USB.
, . , . , , . , . . . , remove.bg .
, remove.bg, . , , , . , . .
. - . , , . , , . , , , .., .
SaaS
? . – , . , .
? . , , . – PayPal Fondy , Best2Pay .
API. , API API .
? , , . , , , .
: , . . , - .
«». -, benzin c , . -, . -, io. -, – , , . -, .
, , benzin . , .
? WordPress + WooCommerce. WordPress, , . , PayPal, Fondy Best2Pay, «» «» . WordPress .
, remove.bg. . . , . , . -, . -, . -, - .
, , , :
, , :
, , "":
benzin neural network, , , remove bg .
? . , , . , . , "HD" . ?
, . . , , , , ?
. , . , , . . , . , ( , ) .
, benzin.io . .
- , . .
– WordPress Woo Commerce, https://benzin.io , .
- , API. PayPal, Fondy Best2Pay. . .
– , .
– , , , , .
– , , , , 28 4-5 . .
– , . Figma.
– , .
– , ( ).
– WP, , , .
– , iOS Android.
. , , . , , , . . , , . , . , - . iOS Android, API. Flutter, .
, 2021 :
: 2 100,
- .. 65%
- .. 28%
- .. 7%
: 2 760
: 6 580
: 12,5%
: 95%
: 73%
: 2 000
: 17%
: 1,3%
: 8,5$
: 460
Figma, Figma 300 . , WordPress+Woo Commerce Shopify .
, , benzin.io – . - . . , . 60 . . , , .
, , . :
Google, – remove.bg. , , . 29 490 000 . 30 3 . 1/1000 .
- : ?
- : , .
- : , .
: , .
, : , - . . , , . , , , , . 1,5% . - , .
:
( )
( )
, ( 1% )
, , 300 000$ 500 000$ . , / 1%. , 1% 29 . 290 , 1$ ( ), 290 . , , .
remove.bg . 4 2020 . 333 . , . , , . 8$.
- -, , , .
, - . , : , . , , 200 000 . , 100 , 100 . .
, Photoshop, Woo Commerce, Shopify, - -. , . – . , , .
, 100 , :
,
.
, 100$ 10 , :
200%
20%
27% 52%
10
10 , 30% .
, . . , c 600$, 6000$. , . ? , . , . , , API . , , . , , , , .
-. . WordPress + Woo Commerce, Shopify, , - .
? , , remove.bg. , - . .
, - , . - , - «» 300-500 .
, . , , 100 , -, . - .
, Canon Nikon 11.4 . - (Canon & Nikon). .
, - , : BENZIN neural network - .