Abstract:

Industry (

Because video surveillance can record and restore the monitored scene, in recent years, video surveillance gradually from professional applications to various civil and household fields, various camera manufacturers have also launched a variety of models of cameras and solutions. Due to the large amount of video surveillance data, various problems will also be encountered:

  1. Storage resources consume a lot and grow rapidly. Video monitoring business extreme storage resources, play a scale of Internet video monitoring services often need to petabytes even more storage resources, common IDC or xiao yun service providers limited infrastructure, it’s easy to become the bottleneck of cloud outbreak of video monitoring business growth, once appear, storage bottlenecks, expanding is difficult and high cost of migration.
  2. High requirements on data reliability. Surveillance video data is the core value of security monitoring services and must not be lost. Especially in the Internet business model, if the video data can not be lost, security monitoring service providers can provide their end users with reliable video storage services.
  3. Bandwidth resources are consumed, and upstream and downstream traffic is asymmetric. 7*24 hours video surveillance services that write video streams are asset-heavy services. Tens of thousands of cameras and several Gigabytes of concurrent upload traffic bring huge bandwidth costs. In addition, the video surveillance business has the obvious characteristics that the upstream traffic is much larger than the downstream traffic, and in the Internet business model, the downstream traffic is the effective business traffic of the industry. In traditional IDC architecture, traffic asymmetry directly leads to the increase of invalid cost.
  4. Users’ viewing experience is difficult to guarantee under massive concurrent services. Fluency is the most important user experience for video surveillance. Tens of thousands of videos are concurrently written and watched, requiring high-quality infrastructure and strong product performance to ensure the quality of video experience. Video streams are generally stored in fragments. If any fragment fails to be written, the playback will be delayed.
Ali Cloud provides video surveillance solutions based on Ali cloud storage service to solve the above problems faced by security monitoring services under the traditional IT architecture. Since the launch of video surveillance storage and playback solutions in 2015, it has helped a large number of video surveillance enterprises to solve the problems encountered in the process of cloud access. For different video surveillance manufacturers, different solutions have also been launched:

Solution 1 for traditional video surveillance platform:

Reference Framework:

Plan interpretation:

  1. The video stream is pushed by the camera to the streaming media server (users need to build their own ECS).
  2. The streaming media server performs TS slices of video streams (which need to be built by users using ECS) and transfers them to OSS.
  3. The video broadcast service can be directly implemented by OSS or forwarded by streaming media servers.
  4. Video Playback services can be directly pulled from the OSS, or the streaming media server can be pulled from the OSS and forwarded to the client

Features and advantages of the solution:

  1. Elastic calculation, so that the single camera comprehensive cost down 30%, the overall calculation cost down 50%. Low-cost and low-configuration cloud servers are used to ensure service continuity, and elastic scaling services are used to automatically increase the number of cloud servers in peak hours to ensure normal service running. At the same time, the elastic scalability can improve resource utilization. Moreover, zero uplink bandwidth charges further reduce the cost of surveillance video uploading, making the overall cost structure more reasonable.
  2. A smoother video viewing experience. Ali Cloud high quality BGP bandwidth brings lower playback delay; Ali Cloud object storage has a very low write failure rate, which can effectively avoid the phenomenon of slow video viewing due to the failure of writing video file slices.
  3. Massive video file storage capability. Ali Cloud object storage (OSS) can provide massive file storage, with 10 9 data reliability, and provide in-city hot backup and remote disaster recovery solutions, so that massive surveillance videos can be permanently stored and available at any time.
  4. Better scalability. Storage, computing, and distribution in the solution are easily scalable.

Solution 2: Camera solution

Reference Framework:

Plan interpretation:

  1. After logging in to the video source (camera or streaming media server) and obtaining the permission to access OSS (Ali Cloud provides authentication service to access OSS), the video stream will be directly pushed to the Object storage OSS of Ali Cloud.
  2. Users use aliyun’s unique add-on write function to save the video code stream to OSS.
  3. If the user saves the video in HLS (TS+M3U8) format, which is one of the most common video file formats on the Internet, the user can view the video stream and playback in real time through mobile phones and computers without using other services. Perfect cloud video real-time viewing and video playback cloud.

Scheme advantages:

  1. Abundant access bandwidth ensures the smooth operation of the whole service even if there is a large number of sudden access. 2. Smoother video viewing experience. Ali Cloud high quality BGP bandwidth brings lower playback delay; Ali Cloud object storage has a very low write failure rate, which can effectively avoid the phenomenon of slow video viewing due to the failure of writing video file slices.

3. Massive video file storage capability. Ali Cloud object storage (OSS) can provide massive file storage, with 10 9 data reliability, and provide in-city hot backup and remote disaster recovery solutions, so that massive surveillance videos can be permanently stored and available at any time.

  1. Better scalability. Storage, computing, and distribution in the solution are easily scalable.
  2. File life cycle management. For the scenario where a large number of expired files need to be deleted periodically, ali Cloud OSS storage service supports automatic deletion of expired files, eliminating the worry of deleting expired files.
  3. Disaster recovery capability. Ali Cloud OSS provides multi-node data replication to easily realize remote data disaster recovery.
  4. Industry technical ability. Video surveillance camera heads all over the world are used in Aliyun’s services. Aliyun has accumulated a large amount of industry and technical experience in the field of video surveillance, which can ensure the smooth operation of users’ business.

Scheme 3 for cameras that support RTMP push stream protocol:

Reference Framework:

Plan interpretation:

  1. The OSS implementation receives RTMP streams directly and converts them into HLS for storage in OSS, and implements on-demand and live streaming through downstream traffic of OSS. 2. Users need to apply for a bucket on the OSS to enable the live streaming service.

3. Convert live data to HLS (TS+M3U8) and save it as an OSS file. You can choose whether to retain the file for voD.

  1. HLS (TS+M3U8) is the format of live video and on-demand. The customer obtains the M3U8 address and plays it directly.

Scheme advantage: except the advantages of scheme ONE and scheme two. There are also the following characteristics:

  1. The system architecture is simple. As long as the camera supports RTMP protocol, the cloud storage can be realized without any modification of camera hardware and software.
  2. Support HLS live broadcast and on-demand functions, easy to achieve Internet applications;

In a word, compared with traditional storage solutions, Aliyun video surveillance solution solves the following problems for the high availability of video surveillance data: 1. High requirement on write success rate Because the failure of video data writing and storage will lead to the failure of watching or even playing, ali Cloud storage is specially optimized for the scene of massive file writing in the video surveillance industry, providing the industry’s top write success rate, far ahead of the peer, and has been tested by a large number of customers’ massive data. 2. High data reliability Requirements The cloud storage system automatically implements three-digit data backup and provides high data reliability with 11 nines, ensuring the security and reliability of video files. Support access of millions of video devices, flexible and reliable architecture, flexible expansion, can be opened and released according to different business needs. For an extreme example, if someone rushed into the room to take away some of ali cloud server, also don’t have to worry about the data will be leaked or lost, because the data is stored encrypted file fragments, don’t see any a complete file, while three backup camera photograph content can be safe in the cloud back second level recovery. Truly achieve the data can not see, touch, take away.

Mining the industry value of monitoring data in the cloud

A, real-time video analysis —- camera meets cloud intelligent video analysis, from see clearly to see understand

After the video is uploaded to the cloud, ali Cloud AI service will be notified through the message service. After obtaining the video, specific video analysis will be performed and the video analysis results will be notified to users in real time. The fully intelligent cloud video analysis system can effectively avoid information omission, so that all kinds of high risks can be immediately notified to the processing personnel at the beginning of occurrence, so as to avoid the occurrence of events. Greatly improve the effect and quality of the whole video monitoring system, reduce the intensity of the staff.

• Face recognition: family face smart time album, kindergarten baby smart classification video, smart doorbell.. • Number of people statistics: shopping and supermarket flow statistics, hot area analysis, cashier queue analysis… • Specific behavior analysis: family elderly fall recognition, gesture recognition, baby crying recognition, intrusion detection… • Intelligent transportation: license plate recognition, congestion detection, traffic violation recognition • Public protection: area protection, intrusion detection… • Public security investigation: suspect identification, group events • Kindergarten: baby abnormal behavior identification, crying persistence, fall detection…

Second, offline video analysis —- intelligent mining of stock video gold mine, so that data play infinite value

After video files are uploaded to the cloud, a large number of video files occupy a large amount of storage space. Due to cost and other factors, video data is usually deleted periodically after being stored for a certain period of time. So this huge amount of data doesn’t produce any value, and the effect is that storage is money and not money. Actually? It’s a video data gold mine! By ali cloud or video analysis algorithm research, not only can the value of stock video, you can even cross-border development of all kinds of business, video to concentrate: remove duplicate the background, will be meaningful in the video scene automatically stored permanently, video structured tags: unstructured video content understanding into structured data, intelligent transportation: Offline violation detection, secondary data mining of massive public security, bayonet and other non-traffic camera data, punishment of illegal vehicles • Public security criminal investigation: massive video search, case clue association…

Click to view the original article