Traffic Light Recognition using Deep Learning

Traffic Light Recognition using Deep Learning

Vamshi Gudavarthi

Vamshi Gudavarthi

Lakeside, California

Recognizing Traffic lights in the driving direction using Deep Learning. The dataset contains images taken by the drivers using Nexar app.

Artificial Intelligence

  • 0 Collaborators




This project is pursued as a final project for the course Neural Networks and Pattern Recognition taken at University of California, San Diego. We tried to solve the problem posed in Nexar challenge shown in the following link, ( Although the challenge is over by then we are able to solve to the extent we were almost there matching the 2nd rank shown in their leaderboard. Also we localized the region of traffic lights using UCSD Traffic lights dataset where data for bounding boxes is also available. Apart from this I had many other machine learning and deep learning projects to my credit. Also I am currently enrolled for graduate study at UCSD specializing in Intelligent Systems, Robotics & Control. Following is one of the reference we used for our project (Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks by Shaoqing Ren, Kaiming He, Ross Girshick, Jian Sun)

Default user avatar 57012e2942

Siddharth N. created project Gesture Recognition System for soldiers

Medium d177d58c e0e0 4930 a7fd 546020c23721

Gesture Recognition System for soldiers

It is a gesture classification system for soldiers where cameras cannot be used.The system I made can classify 40 of the standard army gesture.It uses a support vector machine to classify the gestures.I created my own dataset for training. The list of gestures contain static as well as dynamic gestures. Different algorithms in terms of number of features was used for classifying them. The list of gestures can be found here: Further improvement will be made using wireless modules and increasing mobility.

Medium img 20171008 033022

Burhan K. updated status

Medium img 20171008 033022

Burhan KAYA

AI , Cyber Security, IOT

Hello I'm an Electrical and Electronics Engineer. I do research on artificial intelligence and IOT. Besides this, I have a project with artificial intelligence algorithms. Therefore, using MODIVUS will contribute to me. I do not want to be deprived of such an impossibility. I'm glad you helped me.

Medium profilev2

Chaplin M. updated status

Medium profilev2

Chaplin Marchais

Well its now 7am and I have been up for about 30 hours.... but the image recognition is now actually working in azure!! Now time to take a power nap and then do some optimization with Intel's awesome suite of tools!

Medium profilev2

Chaplin M. updated status

Medium profilev2

Chaplin Marchais

Fifth night in a row I find myself still in front of the computer at 2AM.... I think I got up at-least twice today though!! Progress... Oh well, the future doesn't build itself! ...... yet....

Medium mol

Moloti N. created project Intelligent Home Security: Africa Motion Content encoder decoder using Deep Neural Networks

Medium 18657e03 c017 4a3b b485 57589b45e7a5

Intelligent Home Security: Africa Motion Content encoder decoder using Deep Neural Networks

We propose the use of Drones to help communities enhance their security initiatives, to identify criminals during the day and at night. We use multiple sensors and computer vision algorithms to be able to recognize/detect motion and content in real-time, then automatically send messages to community members cell phones about the criminal activities. Hence, community members may be able to stop house breakings before they even occur.

Machine Intelligence Algorithm Design Methodology


We propose a deep neural network for the prediction of future frames in natural video sequences using CPU. To effectively handle complex evolution of pixels in videos, we propose to decompose the motion and content, two key components generating dynamics in videos. The model is built upon the Encoder-Decoder Convolutional Neural Network and Convolutional LSTM for pixel-level prediction, which independently capture the spatial layout of an image and the corresponding temporal dynamics. By independently modeling motion and content, predicting the next frame reduces to converting the extracted content features into the next frame content by the identified motion features, which simplifies the task of prediction. The model we aim to build should be end-to-end trainable over multiple time steps, and naturally learns to decompose motion and content without separate training. We evaluate the proposed network architecture on human AVA and UCF-101 datasets. We show state-of-the art performance in comparison to recent approaches. This is an end-to-end trainable network architecture running on the CPU with motion and content separation to model the spatio-temporal dynamics for pixel-level future prediction in natural videos.

// We then use this AMCnet pretrained model on the Video feed from the DJI Spark drone, integrated with the Movidius NCS to accelerate real-time object detection neural networks.

See More

No users to show at the moment.

Bigger bob duffy 3d head avatar
  • Projects 4
  • Followers 130

Bob Duffy

Folsom, CA, USA