Real Time Vehicle Detection Using YOLO

This is a project for Udacity self-driving car Nanodegree program. The aim of this project is to detect the vehicles in a dash camera video. We use the You Only Look Once (YOLO) to detect the vehicles from a dash camera video stream, and is able to achieve almost real time (35FPS) processing.

The code for the project is The final video output is here.Introduction to object detectionDetecting vehicles in a video stream is an object detection problem. An object detection problem can be approached as either a classification problem or a regression problem.

In the classification approach, the image are divided into small patches, each of which will be run through a classifier to determine whether there are objects in the patch. The bounding boxes will be assigned to patches with positive classification results. In the regression approach, the whole image will be run through a convolutional neural network directly to generate one or more bounding boxes for objects in the images.

In this project, we will implement the version 1 of tiny-YOLO in Keras, since its easy to implement and are reasonably fast.The tiny-YOLO v1The YOLO approach of the object detection is consists of two parts: the neural network part that predicts a vector from an image, and the postprocessing part that interpolates the vector as boxes coordinates and class probabilities. For the neural network, the tiny YOLO v1 is consist of 9 convolution layers and 3 full connected layers.

Each convolution layer consists of convolution, leaky relu and max pooling operations. The first 9 convolution layers can be understood as the feature extractor, whereas the last three fully-connected layers can be understood as the regression head that predicts the bounding boxes. There are a total of 45,089,374 parameters in the model.

The output of this network is a 1470 vector, which contains the information for the predicted bounding boxes. The information is organized in the following wayThe 1470 vector output is divided into three parts, giving the probability, confidence and box coordinates. Each of these three parts is also further divided into 49 small regions, corresponding to the predictions at the 49 cells that form the original image.

In postprocessing steps, we take this 1470 vector output from the network to generate the boxes that with a probability higher than a certain threshold.Training the YOLO network is very time consuming. We use the pretrained weights from here (172M) and load them into our Keras model.

ResultsThe following shows the results for several test images. We can see that the cars are detected correctly:Here is the result of applying the same pipeline to a video. In order to see the true performance of the model, each frame is processed independently and no temporal smoothing is applied.

We see that the detection is fairly accurate, with just one false positive.The YOLO is known to be fast. In this project, the video is processed on a Nvidia 1080Ti GPU and is able to achieve 35FPS without batch processing.

AcknowledgmentThanks to allanzelener and Trieu for their helpful discussion on YOLO. Thanks to Calvenn Tsuu for performance testing.ReferenceJ.

Redmon, S. Divvala, R. Girshick, and A.

Farhadi, You Only Look Once: Unified, Real-Time Object Detection, arXiv:1506. 02640 (2015). J.

Redmon and A. Farhadi, YOLO9000: Better, Faster, Stronger, arXiv:1612.08242 (2016).

darkflow, RELATED QUESTION Is there any career or scope for Bio-Medical Engineering in North India? As the OP didn't specify any particular thing and as OP appears from India.nI assumed and edited the question to say that its North India, and corrected spelling of Career too.

Career or Scope for Biomedical Engineering in North India. nI hail from North and have experienced Biomedical Engineering there, So what are your options for the samenCareer wise: Biomedical Engineering Jobs are available in Hospitals in Delhi NCR, Chandigarh as well as other important tier-2 cities of North. Apart from that there are n number of companies in Delhi NCR for Biomedical Engineering.

Delhi holds one of the largest surgical instruments companies in India. These companies do manufacturing as well as trading of medical equipment. Apart from that there are many companies which employ engineers for different profiles like application specialists, Service and sales engineer.

Most of my friends in Delhi- NCR are well placed and growing decently in their careers. Research wise: Center for Biomedical Engineering, AIIMS and IIT Delhi is one of the most forward research center in Biomedical Engineering in the country. They also have collaboration with Stanford for Stanford India BiodesignnApart from this we have National Brain Research Center in Gurgaon, doing amazing research in BiomedicalnDRDO lab in delhi does a lot of work in Biomedical EngineeringnDTU also has researchers working in Biomedical EngineeringnIn Chandigarh we have CSIO, CSIR labs which have premier facilities for Biomedical ResearchnIIT Ropar also offers Biomedical Research along with NIT Jalandhar which has a stronghold in ResearchnBiomedical Engineering as a Masters level course offered in ITBHU also offers good research opportunities, IIIT alahabad also has good opportunities along with NIT AllahabadnnThere is a lot of scope in Biomedical Engineering in upcoming years not just NORTH INDIA but all over India.

The Foreign investments on Medical Device research are increasing like anythingnSo Dont Lose hope, Focus on your Basics, We are on a ROLL. .Is there any career or scope for Bio-Medical Engineering in North India?

Is their any career or scope for Bio-Medical Engineering in North India?