معرفي دادگان ايرپلاك براي بازشناسي پلاك خودروهاي ايراني

Introducing a new dataset (Iranian Plate) for Iranian licence plate recognition

ﺷﻨﺎﺳﺎﯾﯽ ﭘﻼك ﺧﻮدرو ﯾﮏ ﺟﺰء اﺳﺎﺳﯽ از ﺳﯿﺴﺘﻢ ﻫﺎي ﻣﺨﺘﻠﻒ ﺣﻤﻞ و ﻧﻘﻞ ﻫﻮﺷﻤﻨﺪ اﺳﺖ ﮐﻪ ﻫﻤﯿﺸﻪ اﻧﺘﻈﺎر ﻣـﯽ رود دﻗﯿـﻖ و ﮐﺎرآﻣـﺪ اﻧﺠﺎم ﮔﯿﺮد. رﺷﺪ ﺑﯽ روﯾﻪ ﺗﻌﺪاد ﺧﻮدرو ﻫﺎ، ﻣﻨﺠﺮ ﺑﻪ ﺑـﺮوز ﻣﺸـﮑﻼت زﯾـﺎدي ﺑـﺮاي ﺗﺸـﺨﯿﺺ ﺧـﻮدرو در زﻣﯿﻨـﻪ ﻫـﺎي ﻣﺨﺘﻠـﻒ از ﺟﻤﻠـﻪ ﮐﻨﺘﺮل ﺗﺮاﻓﯿﮏ، ﻣﺪﯾﺮﯾﺖ ﭘﺎرﮐﯿﻨﮓ، ﻋﻮارﺿﯽ ﺑﺰرﮔﺮاه ﻫﺎ و ﻏﯿﺮه ﺷﺪه اﺳﺖ. ﺑﺎ اﺳﺘﻔﺎده از ﯾﮏ ﺳﯿﺴﺘﻢ ﺧﻮدﮐﺎر ﺟﻬﺖ ﺗﺸﺨﯿﺺ ﺷـﻤﺎره ﭘﻼك ﺧﻮدرو، ﻣﯽ ﺗﻮان ﺗﺎ ﺣﺪود زﯾﺎدي ﺑﺮ اﯾﻦ ﻣﺸﮑﻼت ﻓﺎﺋﻖ آﻣﺪ. ﯾﮑﯽ از ﺑﻬﺘﺮﯾﻦ روش ﻫﺎي اراﺋﻪ ﺷﺪه ﮐـﻪ ﺗـﺎ ﺑـﻪ اﻣـﺮوز ﮐـﺎراﯾﯽ ﺑﺴـﯿﺎر ﺑﺎﻻﯾﯽ در ﺗﺸﺨﯿﺺ ﭘﻼك دارد، اﺳﺘﻔﺎده از ﺷﺒﮑﻪ ﻫﺎي ﻋﺼﺒﯽ ﭘﯿﭽﺸﯽ )CNN( اﺳﺖ. در اﯾﻦ ﭘﮋوﻫﺶ ﺳﻌﯽ ﺑﺮ اﯾﻦ اﺳﺖ ﮐﻪ ﮐﺎراﯾﯽ اﯾﻦ ﺷﺒﮑﻪ ﻫﺎ ﺑﺮاي ﺗﺸﺨﯿﺺ ﭘﻼك ﺧﻮدروﻫﺎي اﯾﺮاﻧﯽ روي دادﮔﺎﻧﯽ ﺟﻤﻊ آوري ﺷﺪه در اﯾﻦ ﻣﻘﺎﻟـﻪ، ﻣـﻮرد ارزﯾـﺎﺑﯽ ﻗـﺮار ﮔﯿـﺮد ﻃﻮري ﮐـﻪ در ﻧﻬﺎﯾﺖ ﺑﺘﻮان روﺷﯽ ﺧﻮدﮐﺎر ﺑﺮاي ﺗﺸﺨﯿﺺ و ﺑﺎزﺷﻨﺎﺳﯽ ﭘﻼك ﺧﻮدروﻫﺎي ﻓﺎرﺳﯽ اراﺋﻪ ﺷﻮد. در روش ﭘﯿﺸﻨﻬﺎدي اﺑﺘﺪا ﭘﻼك ﺧـﻮدرو ﻣﮑﺎن ﯾﺎﺑﯽ ﻣﯽ ﺷﻮد، ﺳﭙﺲ ﺣﺮوف و اﻋﺪاد آن ﺟﺪاﺳﺎزي و اﺳـﺘﺨﺮاج ﻣﯽ ﺷـﻮﻧﺪ و در ﻧﻬﺎﯾـﺖ ﺑـﺎ اﺳـﺘﻔﺎده از ﺷـﺒﮑﮥ ﭘﯿﭽﺸـﯽ اﻟﮑـﺲ ﻧـﺖ ﻧﻮﯾﺴﻪ ﻫﺎ دﺳﺘﻪ ﺑﻨﺪي ﻣﯽ ﺷﻮﻧﺪ. از ﻧﻮآوري ﻫﺎي ﭘﮋوﻫﺶ ﻣﯽ ﺗﻮان ﺑﻪ ﺟﻤﻊ آوري دادﮔﺎﻧﯽ ﺑﺎ 5000 ﺗﺼﻮﯾﺮ از ﺧﻮدروﻫـﺎي اﯾﺮاﻧـﯽ، اﺳـﺘﻔﺎده از اﻧﺘﻘﺎل ﯾﺎدﮔﯿﺮي )Transfer Learning( ﺑﺮاي ﺗﺸﺨﯿﺺ ﭘﻼك ﺧﻮدروﻫﺎي اﯾﺮاﻧﯽ و ﺑﺎزﺷﻨﺎﺳﯽ ﺗﻮاﻣﺎن ﺣﺮوف و ارﻗﺎم ﻓﺎرﺳﯽ اﺷﺎره ﮐﺮد. ﮐﺎراﯾﯽ روش ﭘﯿﺸﻨﻬﺎدي روي دادﮔﺎن ﺟﻤـﻊ آوري ﺷـﺪه ﺑـﺎ دﻗـﺖ 70 درﺻـﺪ داده ﻫـﺎي آﻣـﻮزش و 30 درﺻـﺪ داده ﻫـﺎي آزﻣـﻮن ﺑـﻪ دﻗـﺖ 98/20 درﺻﺪ رﺳﯿﺪه اﺳﺖ.

License Plate Detection is an essential part of smart transportation systems which is always expected to be accurate and efficient. The exponential growth of the number of vehicles has led to many problems for vehicle detection in different fields such as traffic control, parking lot, toll highways and so. Using an automatic system for plate number detection of the vehicle, a large number of problems can be solved. One of the best methods for overcoming this problem is using a convolutional neural network approach. They have shown impressive performance in various computer vision tasks. In this work, we tried to introduce a new dataset of Iranian cars' plates. Then using transfer learning and Alexnet, we have proposed a new approach for detection and classifying the plate numbers. We conducted comprehensive experiments on the new dataset. The experimental results show that the proposed method has achieved to 97.35% accuracy.