Traffic Light Recognition using Deep Learning

Traffic Light Recognition using Deep Learning

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

Ashish A. created project Snowball Sampling

Medium b2d7d897 f361 45f7 aef4 ebf4d5695e56

Snowball Sampling

When do we say that a graph has become large or the amount of data in the graph has become big? Why do we sample a graph? When do we say that a graph is sampled? What should be the size of our sampled graph? What difference does it make to work on the original graph and the sampled graph? These are some questions that are very common when people start working on real world of graphs that often span hundreds of millions or even billions of nodes and interactions between them. By the thumb of rule, we can say that 'large graphs' are those graphs exploration of which requires long computation time and 'big data' is typically the data which takes too much memory space to be stored on a single hard drive. Why do we need to sample the original graph? First and the foremost reason is that the sheer size of many networks makes it computationally infeasible to study the entire network. Moreover, the size of the network may not be as large but the measurements required to observe the underlying network are costly. Thus, network sampling is at the heart and foundation of our study to understand network structure. A good sampled graph must include useful knowledge. Our primary goal is to find the important properties that effectively summarizes the graph.

Graphs are used to represent real life situations where entities of internet are related to each other. In such situations, entities can be represented as nodes, and the relationship between them can be represented as edges. Graph Modelling of real life situations results into into networks. Thus, there are transport networks, road networks, biological networks, technology networks etc. Analysis and importance of these networks has given rise to a recent discipline of network science. Analysis of networks that are large and dense is a challenging task because of the associated computational expense. Focusing on smaller and dense areas of network is often preferred due to two reasons : \newline (i) reduced computation in terms of both time and memory (ii) better insights.

Default user avatar

Dheeraj S. updated status

Default user avatar

Dheeraj Sharma

On survey Literate the Peoples and Students regarding all subjects with Modern technology. With Regards to INTEL (Gordon Moore, Robert Noyce) CHIPNXT TECHNOLOGIES PVT LTD

Medium 0

Chandana R. updated status

Medium 0

Chandana Ravindra Prasad

Cross-Domain Sentiment Analysis on Enron Email data Oct 2017 – Dec 2017 Used transfer learning or domain adaptation technique to observe the applications of inter-domain approach. Training data was gathered from three different sources: Yelp, Amazon and IMDb. The model was then tested on Enron Data. This proved to be an efficient technique to apply when the target domain contains limited labelled data and collecting more data would be costly and tedious.

Predicting retweet possibilities based on sentiment analysis of historical tweets Jan 2018 - May 2018 Performing predictive analysis of future tweet traffic for a specified set of users based on sentiment analysis of that user's historical tweets.

Android Application Development Jan 2018 - May 2018 Created many fully functional applications for Android devices like Stock Watch, MultiNotes, Temperature Converter etc. Used AsyncTask, RecyclerView, CardView, SQLite.

Medium 13505895

Farhan A. updated status

Medium 13505895

Farhan Ansari

My interest in the areas of Natural Language Processing and Artificial Intelligence. In a recent project, I will be starting off by implementing a few basic flagging algorithms and then proceed to integrate them with the workflow. The aim is to keep testing and iterating over different techniques throughout the internship period, to find out what works and what doesn't as well. I am excited to know how this project might pan out once strong flagging algorithms are brought into the picture! Also, I'm building some AI based application to make tool smarter and faster in processing the requirements.

See More

No users to show at the moment.