Using amazon recognition to enhance bike safety

Cycling is a happy way to live right, enjoy the environment, and connect with friends and acquaintances. However, riding depends on dangerous, especially in cases where cyclists and cars share the road. According to NHSDA, in the United States the average 883 bicycles were killed in road accidents, at the rate of only 45,000 injuries. While full bike retalities for more than 2% of all travel agabicities in the United States, such as a bicycle river, it is afraid to be extracted outside the road or truck. To better protect, many bicycles begin riding with the previously placed cameras or back to his bicycle. In this blog course, I will show a machine reading solution that cyclists can use to add better-close calls.

Many US countries and countries around the world have some type of legal law. The 3-foot law requires cars to supply 3 meters of distance when a bicycle passes. To elevate the road safety, cyclists increasing their rides, and if they meet a dangerous situation where they do not have the right grade, they can provide a video of local law to help proper performance. However, finding one encounters in a number of hours to ride more hours and often requires skills special videos to produce a brief piece of meetings.

Resolving some of these problems, improving a simple solution I use the Amazon Realusion video. The Amazon's wings can find the labels (actually things) and the time period that the item is found in the video. Amazon's rename can be used quickly to find any cars from the recording video.

If the Cyclist camera record of a passing car, you must see that the car is very close to bicycles – in other words, if the vehicle is within the 3 grade prescribed by law. If so, then I want to produce a meeting clip, which can be given the right authorities. The following figure shows a view from the cyclist camera with a binding boxes that produce a very closer vehicle. The box below the picture shows an estimated area of ​​three feet around the bike.

Looking for everything

The construction of the solution is shown in the following.

The remedies of the solution are:

1. When bicyclist finishes riding, loading their MP4 videos from the shipment to the simple storage service (Amazon S3)
2. Bucket is prepared for notice of an Event of S3 which sends an object performed by the AWS LAMBDA
3. Lambda work kicks AWS steps to work with first work by calling the startlabelled API as part of Amazon Ready Video. This page StartLabelDetection API is prepared to find Bus, Car, Fire Truck, Pickup Truck, Truck, Limobesides Moving Van Like the labels. Ignore other labels without cars like License Plate, Wheel, Tirebeside Car Mirror.
4. Amazon Regenignigning API Returns a JSON set pointing selected labels and Timestamp of the findings.
5. This JSON's outcome is sent to a lambda work to make geometric figures to find that the vehicle's box is full of a safe bicycle.
6. Any cohesion found produced and transferred to AWS Elemental MediaConvert, which may cause snippets of video to meet the meetings available, using CreateJob War
7. MediaConvert creates these videos and loaded the S3 bucket.
8. Another lambda work is called to produce URLs signed videos. This allows videos to be temporarily downloaded by anyone for signed URL.
9. Simple Amazon Notifications Service (Amazon SNS) sends email message with links to signed URLs.

Requirements

To use a solution that has been expressed in this post, you must be:

1. AWS account with appropriate permits to allow you to submit gays AWS AWS
2. Video recording in MP4 format with .mp4 expansion using IH.264 cocec. The video should be from the front or backup camera, in any OFF-theer seller (for example, GOPRO, a DJI, or cycliq. The higher file size is 10 GB.

Using a Solution

1. Use this solution to your area or select Launch the stack. This solution will use at the AWs US East (N. Virginia) in the US-1 AWS region.

2. This page Create a stack The page from Cloudformation Dashboard appears. At the bottom of the page, select Next.
3. Occupile Specify Stack Details Page, enter the email address where you would like to receive notifications. Designate Next.
4. Choose the box comments I agree that AWS Cloudvation can create IAM services and select Next. Designate Place down and the installation will begin. The solution takes 5 minutes to be included.
5. You will receive your Amazon SNS subscription enrollment. You will not receive emails from a solution unless you verify your subscription.
6. After the stack has completed, select Output tab and pay attention to a bucket name listed below Invest.

Using a solution

To check the cure, I have a sample video where I asked the Stunt driver driving a lot to me.

To start video processing, I upload a video to S3 baker (Inctucket from the Exit Tab). Bucket has been enabled to encryption, so below PropertiesI choose Specify an encryption key and select Use the default craftsmen settings. Determine Resolve Starts the process of loading, as shown in the following.

After a moment, the step job begins to process. After a few minutes, you will receive the links with the links in any congestion identified, as shown in the following.

In my story, the two meters expressed. The first meeting was identified, I rode heavily on the park. However, in the second meeting identified, it shows a dangerous combination I have with my stud driver.

If this have been a dangerous danger, the video clip can be provided with the relevant authorities to help alter the behavior and road safety for everyone.

Putting higher property

Because this is a solid solitude, you pay only what you use. With the renaming of Amazon, you pay the video mins. With mediaConvert, you pay regular video mins, which is each minutes of video and repeated video issuers based on used features. The use of a solution for lambade, action activities, and small SNS and may be falling under a free tire for many users.

Clean

Deleting resources made as part of this solution, go to the Cloudformation Consolence, select a stacked stack, and select Erase.

Store

In this example I have shown using the Amazon's video shares method of a unique form. Amazon RelangGinitic is a powerful computer detector that allows you to get information without photos or video without the passing of the building or managing a machine learning model. It is true, the Amazon's revelation can also manage the charges of useful use than I have seen here.

In this example I showed that using Amazon recycling and other non-limited services can produce a flexibility of such a time – can help improve cycling safety. While you might not be a strong bicycle, the displayed solution can be extended in various charges of use and industry. For example, this solution can be expanded to find wildlife in natural cameras or you can use Amazon Rekognition events for broadcasting videos to find people and packages on the security video.

Get started today through recurring amazon persistence for your computer.

About the writer

Mike George Is the construction of the main solutions of Amazon's web services (AWS) based on Salt Lake City, Utah. He is happy to help customers their technical problems. His interests include software engineering, safety, artificial intelligence (AI), and the learning of the machine (ML).