Project details
On-Road Vehicle Make and Model Recognition
Company: FAVA NAJA Iran.
Period:
Duration: 0.9 years (full-time equivalent)
- C++
- OpenCV
- Win32 API
Software Developer in Shiraz University CVPR Lab project,
Challenges:
• Large number of vehicle categories , Managed to get 92% accuracy for different cameras.
• Huge train and test data, 100,000 images
multiple optimizations to keep the needed Train Data in RAM less than 20GB.
• Exploting CPU resource with the help of modern C++ to make Training Phase
algorithm perform faster and complete in less than 10 days.
• Proccessing images from multiple cameras on network using Windows Async IO
features to reach 12 images per second throughput.
• Cuncorrent SVM and Neural Networks algorithms.
About me
Full Stack Developer
My skills
- JavaScript
- C++
- C# Language
- Node.js
- Embedded Systems
- ASP.NET MVC
- Highcharts
- OpenCV
- OpenGL
- ZigBee
- React
- WPF
- Gulp.js
- TypeScript
- Boost C++
Endorse Amin's project
See Amin's profile
All Amin's projects
-
Zigbee Pro stack network with nxp jn516x chips
(click to see)
(16 weeks FTE)
Client: -
Industry: Electronics
- C++
- Embedded Linux
Software Developer in Shiraz University CVPR Lab project,
Challenges:
• Large number of vehicle categories , Managed to get 92% accuracy for different cameras.
• Huge train and test data, 100,000 images
multiple optimizations to keep the needed Train Data in RAM less than 20GB.
• Exploting CPU resource with the help of modern C++ to make Training Phase
algorithm perform faster and complete in less than 10 days.
• Proccessing images from multiple cameras on network using Windows Async IO
features to reach 12 images per second throughput.
• Cuncorrent SVM and Neural Networks algorithms.
-
Vision based intersection monitoring system
(click to see)
(14 weeks FTE)
Client: Iran Shiraz Municipality
Industry: Science
- OpenCV
- WebSockets
Software Developer in Shiraz University CVPR Lab project,
Challenges:
• Large number of vehicle categories , Managed to get 92% accuracy for different cameras.
• Huge train and test data, 100,000 images
multiple optimizations to keep the needed Train Data in RAM less than 20GB.
• Exploting CPU resource with the help of modern C++ to make Training Phase
algorithm perform faster and complete in less than 10 days.
• Proccessing images from multiple cameras on network using Windows Async IO
features to reach 12 images per second throughput.
• Cuncorrent SVM and Neural Networks algorithms.
-
Iran Stock Market Website
(click to see)
(18 weeks FTE)
Client: -
Industry: Computing
- Highcharts
- JavaScript
- Require.js
Software Developer in Shiraz University CVPR Lab project,
Challenges:
• Large number of vehicle categories , Managed to get 92% accuracy for different cameras.
• Huge train and test data, 100,000 images
multiple optimizations to keep the needed Train Data in RAM less than 20GB.
• Exploting CPU resource with the help of modern C++ to make Training Phase
algorithm perform faster and complete in less than 10 days.
• Proccessing images from multiple cameras on network using Windows Async IO
features to reach 12 images per second throughput.
• Cuncorrent SVM and Neural Networks algorithms.
-
On-Road Vehicle Make and Model Recognition
(click to see)
(0.9 years FTE)
Client: FAVA NAJA Iran.
Industry: Science
- C++
- OpenCV
- Win32 API
Software Developer in Shiraz University CVPR Lab project,
Challenges:
• Large number of vehicle categories , Managed to get 92% accuracy for different cameras.
• Huge train and test data, 100,000 images
multiple optimizations to keep the needed Train Data in RAM less than 20GB.
• Exploting CPU resource with the help of modern C++ to make Training Phase
algorithm perform faster and complete in less than 10 days.
• Proccessing images from multiple cameras on network using Windows Async IO
features to reach 12 images per second throughput.
• Cuncorrent SVM and Neural Networks algorithms.
-
Modern Sqlite Cpp
(click to see)
(3 weeks FTE)
Client: Open Source.
- C++
- Standard Template Library (STL)
- Templates
Software Developer in Shiraz University CVPR Lab project,
Challenges:
• Large number of vehicle categories , Managed to get 92% accuracy for different cameras.
• Huge train and test data, 100,000 images
multiple optimizations to keep the needed Train Data in RAM less than 20GB.
• Exploting CPU resource with the help of modern C++ to make Training Phase
algorithm perform faster and complete in less than 10 days.
• Proccessing images from multiple cameras on network using Windows Async IO
features to reach 12 images per second throughput.
• Cuncorrent SVM and Neural Networks algorithms.
-
Installer for an enterprise application
(click to see)
(5 weeks FTE)
Client: -
Industry: Consumer Products
- IIS
- Microsoft SQL Server
- Microsoft Windows Services
Software Developer in Shiraz University CVPR Lab project,
Challenges:
• Large number of vehicle categories , Managed to get 92% accuracy for different cameras.
• Huge train and test data, 100,000 images
multiple optimizations to keep the needed Train Data in RAM less than 20GB.
• Exploting CPU resource with the help of modern C++ to make Training Phase
algorithm perform faster and complete in less than 10 days.
• Proccessing images from multiple cameras on network using Windows Async IO
features to reach 12 images per second throughput.
• Cuncorrent SVM and Neural Networks algorithms.
-
Full-screen web trading app
(click to see)
(7 weeks FTE)
Client: Binary.com
- JavaScript
- WebSockets
- jQuery UI
Software Developer in Shiraz University CVPR Lab project,
Challenges:
• Large number of vehicle categories , Managed to get 92% accuracy for different cameras.
• Huge train and test data, 100,000 images
multiple optimizations to keep the needed Train Data in RAM less than 20GB.
• Exploting CPU resource with the help of modern C++ to make Training Phase
algorithm perform faster and complete in less than 10 days.
• Proccessing images from multiple cameras on network using Windows Async IO
features to reach 12 images per second throughput.
• Cuncorrent SVM and Neural Networks algorithms.
