(For USM Staff/Student Only)

EngLib USM > Ω School of Electrical & Electronic Engineering >

Object recognition on raspberry pi using tensorflow

Object recognition on raspberry pi using tensorflow / Tan Hang Yan
Pembelajaran Mendalam sudah digunakan dengan meluas dalam kebanyakan aplikasi terkini dan telah dilaksanakan dalam sistem terbenam. Terdapat banyak aplikasi dalam Pembelajaran Mendalam akan dilaksanakan kepada masyarakat kita pada masa yang akan datang seperti kereta tanpa pemandu dan penjanaan teks secara automatic. TensorFlow adalah salah satu daripada sumber terbuka kerangka Pembelajaran Mendalam boleh digunakan untuk pengiraan kerangka berprestasi tinggi dan seni bina yang fleksibel memudahkan pengiraan merentasi pelbagai platform. Analisis perlu dilakukan untuk menilai ketepatan dan kepantasan model TensorFlow SSDMobileNet V2 and SSD Inception V2 diaplikasikan dalam pengesanan objek menggunakan Raspberry Pi 3 Model B+ dengan modal, sumber dan masa terhad. Dalam kajian ini, COCO 2017 digunakan untuk menilai ketepatan SSDLite MobileNet V2 and SSD Inception V2 dari segi tahap keyakinan 0.7 untuk mengesan lima kelas objek. Selain itu, input video yang sama digunakan untuk menilai kepantasan SSDLite MobileNet V2 and SSD Inception V2 dari segi bingkai sesaat(FPS). Hasil penilaian ketepatan menunjukkan bahawa SSDLite MobileNet V2 menunjukkan markah pengesanan 69.83% dengan sisihan piawai 3.96% untuk semua lima kelas. Sementara itu, SSD Inception V2 menunjukkan markah pengesanan 73.72% dengan sisihan piawai 4.70% untuk mengesan lima kelas objek. Di samping itu, keputusan penilaian kepantasan menunjukkan bahawa purata FPS SSDLite MobileNet V2 adalah 1.01 dengan sisihan piawai 0.107. SSD Inception V2 memberikan hasil penilaian kelajuan dengan purata FPS sebanyak 0.48 dengan sisihan piawai 0.041. Kesimpulannya, SSD Inception V2 lebih baik berbanding dengan SSDLite MobileNet V2 dalam penilaian ketepatan manakala SSDLite MobileNet V2 lebih baik berbanding dengan SSD Inception V2 dalam penilaian kepantasan. _______________________________________________________________________________________________________ Deep Learning (DL) has been widely used in many applications and implemented in commercial and industrial embedded systems. There are a lot of DL applications that will rule the world in the future such as self-driving cars, healthcare, voice search assistants and automatic text generation. TensorFlow is one of the open source DL frameworks. TensorFlow can be used for high performance numerical computations and its flexible architecture allows easy deployment of computation across a variety of platforms. In this project, the accuracy and speed of TensorFlow pretrained models SSDLite MobileNet V2 and SSD Inception V2 implemented in the developed object recognition system on Raspberry Pi 3 Model B+ are evaluated and compared. In this research, COCO 2017 has been used to evaluate the accuracy of SSDLite MobileNet V2 and SSD Inception V2 in terms of 0.7 confidence level on detecting five classes of object. Furthermore, a video input is used to evaluate the speed of SSDLite MobileNet V2 and SSD Inception V2 in terms of Frame per Second (FPS). The accuracy evaluation result shows that the SSDLite MobileNet V2 has a detection score of 69.83% with standard deviation of 3.96% for all five classes. Meanwhile, SSD Inception V2 shows a detection score of 73.72% with a standard deviation of 4.70% for detection of five classes of objects. In addition, the speed evaluation result shows that mean FPS of SSDLite MobileNet V2 is 1.01 with a standard deviation of 0.107. SSD Inception V2 gives a speed evaluation result with mean FPS equal to 0.48 with a standard deviation of 0.041. With these results, it is concluded that SSD Inception V2 performs better than SSDLite MobileNet V2 in the accuracy assessment while SSDLite MobileNet V2 performs better than SSD Inception V2 in the speed assessment.
Contributor(s):
Tan Hang Yan - Author
Primary Item Type:
Final Year Project
Identifiers:
Accession Number : 875008592
Language:
English
Subject Keywords:
(DL); embedded; systems
First presented to the public:
6/1/2019
Original Publication Date:
3/3/2020
Previously Published By:
Universiti Sains Malaysia
Place Of Publication:
School of Electrical & Electronic Engineering
Citation:
Extents:
Number of Pages - 101
License Grantor / Date Granted:
  / ( View License )
Date Deposited
2020-03-03 11:48:23.778
Submitter:
Mohd Jasnizam Mohd Salleh

All Versions

Thumbnail Name Version Created Date
Object recognition on raspberry pi using tensorflow1 2020-03-03 11:48:23.778