├── benchmark.jpg
└── README.md


/benchmark.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/KBNLresearch/bb_recog/master/benchmark.jpg


--------------------------------------------------------------------------------
/README.md:
--------------------------------------------------------------------------------
 1 | <h1>Book back recognition</h1>
 2 | 
 3 | <h2>Challange</h2>
 4 | 
 5 | 0) Upload image to webinterface.
 6 | 
 7 | 1) Apply Houg (Or simular edge detecting algorithm) & extract segments.
 8 | 
 9 | 2) OCR text per segment (Non commercial OCR-software only).
10 | 
11 | 3) Validate OCR output using ngrams (Or other local service).
12 | 
13 | 4) Lookup ISBN per book (Using service of choice).
14 | 
15 | <h2>Optional:</h2>
16 | 
17 | 5) What is the main catagory of the shelf.
18 | 
19 | 6) What if the shelf would have extra space for x books,
20 |    and what is the shelf is too full, which x books to remove.
21 |    
22 | 7) Think about possibilities when having access to full text versions of the physical books.
23 | 
24 | <h2>Benchmark image:</h2>
25 | 
26 | ![Benchmark](https://raw.githubusercontent.com/KBNLresearch/bb_recog/master/benchmark.jpg)
27 | 
28 | 
29 | 37 Segments.
30 | 
31 | 1) Spiegel van Noord-West Nederland.
32 | 1b) NL 76 D 6000 01
33 | ISBN:  9066185163
34 | 
35 | 2) Spiegel van Zuid-West Nederland.
36 | 2b) NL ..
37 | 
38 | ..
39 | 
40 | ..
41 | 
42 | ..
43 | 
44 | 
45 | 


--------------------------------------------------------------------------------