In this paper, we explain our Traffic Detection technique using OpenCV concept, Neural Networks, Tensorflow, and how it is successfully detecting and identifying vehicles and other roadside attributes such as pedestrians, signs, and lane markings for a thorough analysis through a road surveillance camera image. Our pre-trained SVM model is highly efficient and accurate in performing the desired task successfully. The paper should be of interest to readers in the areas of Image processing, neural networks, and machine learning.