python .\server.py .\result.jpeg.
first run the server and then on argv[1] enter the name of file which will be output of input image provided by client.
python .\client.py .\img3.jpeg.
then clien will connect with the server and on argv[1] enter the image file name of which text is to be recognized.
PS D:\socket programming> python .\server.py .\result.jpeg Server is running Server is ready to accept data ... connected with client having ('192.168.1.101', 59346) CUDA not available - defaulting to CPU. Note: This module is much faster with a GPU. PS D:\socket programming>
PS D:\socket programming> python .\client.py .\img3.png
client conected
Result of text Recognition is:
Do not resolve addresses to hostnames Maximum number 0f hops to search for target Loose source route along host-list (IPv4-only) Wait timeout milliseconds for each reply _ Trace round-trip path (IPv6-only) Source address to use (IPv6-only). Force using IPv4 . Force using IPv6 _
Result in Table form
| Text | Accuracy |
|---|---|
| Do | 99.99 |
| not | 90.46 |
| resolve | 99.99 |
| addresses | 98.68 |
| to | 99.96 |
| hostnames | 99.91 |
| Maximum | 26.89 |
| number | 100.00 |
| 0f | 99.99 |
| hops | 99.71 |
| to | 90.46 |
| search | 100.00 |
| for | 99.97 |
| target | 97.36 |
| Loose | 73.13 |
| source | 100.00 |
| route | 99.77 |
| along | 99.91 |
| host-list | 70.38 |
| (IPv4-only) | 92.15 |
| Wait | 99.99 |
| timeout | 99.99 |
| milliseconds | 99.98 |
| for | 99.99 |
| each | 60.04 |
| reply | 67.17 |
| _ | 99.34 |
| Trace | 79.04 |
PS D:\socket programming>
Text recognition is greatly dependent the condition of handwriting the contrast, Brightness ratios in image and many more factors. To detect the text very accurately our model should be more robust, and fault tolerant in any condition at the same time our model should be fast. So, with the fast and accurate model, server can send data to client very fast and accurate. So, communication will be faster. With this conclusion summing-up my project.