CN103491379A - Method and device for improving intelligent analysis performance - Google Patents
Method and device for improving intelligent analysis performance Download PDFInfo
- Publication number
- CN103491379A CN103491379A CN201310446301.2A CN201310446301A CN103491379A CN 103491379 A CN103491379 A CN 103491379A CN 201310446301 A CN201310446301 A CN 201310446301A CN 103491379 A CN103491379 A CN 103491379A
- Authority
- CN
- China
- Prior art keywords
- resolution
- video
- gpu
- intellectual analysis
- code flow
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Abstract
The invention provides a method and device for improving intelligent analysis performance. The method for improving the intelligent analysis performance comprises the steps that (1) a GPU which supports a rapid hardware decoding and hardware coding technology carries out hardware decoding on an input compressed encoding video bitstream to form a video bitstream with a first resolution ratio, compression and encoding are carried out on the video bitstream with the first resolution ratio to form a video bitstream with a second resolution ratio, the coded video bitstream with the second resolution ratio is stored on a video memory of the GPU, and the second resolution ratio is smaller than the first resolution ratio; (2) the video bitstream with the second resolution ratio is copied to a memory from the video memory so that a CPU can carry out intelligent analysis on the video bitstream on the memory. According to the technical scheme, performance consumption of the CPU is effectively reduced.
Description
Technical field
The present invention relates to technical field of video monitoring, relate in particular to a kind of method and apparatus that promotes the intellectual analysis performance.
Background technology
Along with the extensive use of video monitoring service, the intellectual analysis of monitor video is more and more ripe, comprises that the virtual line of stumbling detects, amount of exercise detects, human face analysis, and packet loss detection (detects in a zone, a kind of intellectual analysis that object suddenly disappears), intensive detection of personnel etc.But the complete intellectual analysis flow process in monitoring field has not only comprised the intellectual analysis algorithm, it also comprises the decoding performance of video flowing simultaneously.Lifting along with current field of video monitoring video camera resolution, as 1080P resolution is 1920 * 1080, the about 700Mbps of its decoded 1080P yuv data, even the i73770 processor also at full capacity parallel parsing 6 tunnel frame per second be 30 frames, the video code flow of the H264 that resolution is 1080P is very big to cpu performance consumption.
Summary of the invention
In view of this, the invention provides a kind of method that promotes the intellectual analysis performance, the method comprises: the video code flow that steps A, the GPU that supports quick hard decoder and hard-disk cartridge are first resolution to the compressed encoded video code stream hard decoder of input, and the video code flow that is second resolution by the video code flow compressed encoding of this first resolution, the video code flow of the second resolution after described coding is stored in the video memory of described GPU, and this second resolution is less than first resolution; Step B, the video code flow of described second resolution is copied to internal memory from described video memory, for CPU, the video code flow in this internal memory is carried out to intellectual analysis.
Design based on same, the present invention also provides a kind of device that promotes the intellectual analysis performance, this application of installation is in intellectual analysis equipment, and the device of this lifting intellectual analysis performance has the GPU that supports quick hard decoder and hard-disk cartridge, and this device comprises: code stream modular converter and code stream replication module.The code stream modular converter, for the described GPU of instruction to the compressed encoded video code stream hard decoder of the input video code flow that is first resolution, and the video code flow that this GPU of instruction is second resolution by the video code flow compressed encoding of this first resolution, the video code flow of described second resolution is stored in the video memory of this GPU, and this second resolution is less than first resolution; The code stream replication module, copy to internal memory for the video code flow by second resolution from described video memory, for CPU, the video code flow in this internal memory carried out to intellectual analysis.
Compared with prior art, the present invention utilizes the GPU that supports quick hard decoder and hard-disk cartridge will deposit in the data that decoded high-resolution data in the GPU video memory is encoded into low resolution, by the high-resolution data of small data quantity, from video memory, copy in internal memory, CPU carries out intellectual analysis to the data in this internal memory again, and whole handling process has greatly reduced the performance consumption of CPU.
The accompanying drawing explanation
Fig. 1 is a kind of process chart of intellectual analysis;
Fig. 2 is the building-block of logic of apparatus of the present invention;
Fig. 3 is the process chart of one embodiment of the present invention.
Embodiment
Some intellectual analysis of monitoring trade only needs the image of low resolution just can realize analytical effect preferably at present, as virtual, stumble line, packet loss detection etc., so aforementioned decoded 1080P yuv data directly can be carried out to dwindling of image in yuv space, then transfer to intelligent analysis module and carry out intellectual analysis.Specifically please refer to Fig. 1.Fig. 1 is the flow process that current monitoring trade generally adopts on the intellectual analysis of the virtual line of stumbling.Video source data is obtained the YUV view data that resolution is 1080P after the soft decoding of H264 by CPU.This YUV view data is directly carried out image and is dwindled to process and obtain the image that resolution is 352 * 288 in yuv space by CPU.Data volume narrows down to 20 times of original left and right thus.Carry out virtual mix the line intellectual analysis in, CPU directly carries out intellectual analysis to the CIF image after dwindling.Because the view data of intellectual analysis is less than originally, so reduced the pressure of CPU in intellectual analysis.But the H264 decoding to the 1080P image has consumed a large amount of performance of CPU, has reduced processing speed, so can not significantly promote on the whole the efficiency of intellectual analysis.
Along with aobvious (the core core video card of Intel company's core, be that GPU and CPU are based upon on same core chip) popularization of the Intel Quick Sync Video technology that has (high-speed image synchronously lard speech with literary allusions part technology), the inventor considers directly by this Quick Sync Video technology, to carry out the H264 decoding.This decode procedure belongs to hard decoder, so can quickly original video code flow be converted to the video code flow of extended formatting.Seem so all problems that solved, but the inventor is through carefully analyzing discovery: the video code flow that adopts Quick Sync Video technique decodes to obtain will be stored in the video memory of GPU, and CPU can not access this video memory data are wherein carried out to intellectual analysis; So can consider further the huge data Replica in video memory is carried out to intellectual analysis for CPU in internal memory.But this data volume is larger, so whole reproduction process will consume the more performance of CPU.And even if this reproduction process resource cost is little, but CPU carry out virtual mix the intellectual analysis such as line in, the data of the high definition obtained after the Direct Analysis decoding, also larger to its performance consumption.
The inventor expects utilizing the encoding function of Intel Quick Sync Video technology through further carefully thinking, will deposit in decoded high-resolution data in the GPU video memory and further be encoded into the data of low resolution.Certainly the high-resolution data after coding also is stored in video memory, because data volume is relatively little, so it is also little to copy to the performance of the CPU consumed in internal memory.Below by an embodiment, describing the present invention program in detail.
Please refer to Fig. 2, Fig. 2 provides a kind of device that promotes the intellectual analysis performance, and this application of installation is in intellectual analysis equipment (intellectual analysis equipment can be an independent equipment, can be also the intelligent analysis module be integrated on server).The device of this lifting intellectual analysis performance has the GPU that supports Intel Quick Sync Video technology, from this device being divided into lower module in logic: code stream modular converter 101 and code stream replication module 102.
Below with the example in Fig. 3, such scheme is set forth.
Video source input intellectual analysis equipment, the GPU(of this intellectual analysis equipment supports Intel Quick Sync Video) video source of input is carried out to hard decoder.This video source is the video code flow that front end transmits through compressed encoding, and the video source of the 1080P that the process H264 of take here encodes is example, and GPU obtains the video data of 1080P yuv space after adopting the H264 decoding storehouse of Quick Sync Video to be decoded to it.Decoded video data is stored in the video memory of this GPU.
Further, GPU adopts the Mpeg2 code database of Quick Sync Video that it is encoded to the CIF code stream video data of the 1080P of this yuv space.Here it should be noted that the CIF code stream is to be determined by the type of the needed code stream of intellectual analysis.A kind of realization can be inputted by the user code stream resolution needed in software code, then by cpu instruction GPU, carries out relevant treatment.This Mpeg2CIF code stream after Intel Quick Sync Video technology for encoding is stored in the video memory of GPU.Next the CIF code stream of Mpeg2 is copied in the internal memory of CPU, then CPU, to being decoded through the CIF code stream of Mpeg2 compressed encoding in this internal memory, obtains the CIF code stream of yuv space, then it is carried out to the virtual line intellectual analysis of stumbling.What when encoding here, adopt is the Coding Compression Algorithm of Mpeg2, in fact also can use other Coding Compression Algorithm.
By above scheme, can find out: the speed of the GPU hard decoder of support Intel Quick Sync Video is very fast, so this process is consuming time very short; The video flowing that is small data quantity by Intel Quick Sync Video technology for encoding by the video flowing of decoded big data quantity, consuming time also very short; The video flowing of small data quantity copies to internal memory also than comparatively fast from video memory, and CPU is also lower to the decode performance that expends of the video data of low resolution, so the present invention program is very meaningful to promoting on the whole the intellectual analysis performance.Experimental results show that when carrying out the virtual line intellectual analysis of stumbling, the technology of employing Fig. 1 can only be supported the analysis of 6 road H26430 frame 1080P simultaneously; But adopt when of the present invention, can support the virtual line intellectual analysis of stumbling of 18 road H26430 frame 1080P simultaneously.
Above embodiment take the GPU of support Quick Sync Video of Intel Company to describe as example.In fact except the GPU of the support Quick Sync Video of Intel Company, other supports the GPU of quick hard decoder (hard decoder refers to use specialized hardware to be decoded) and hard coded (hard coded refers to use specialized hardware to be encoded) to be applicable to too the present invention, such as support the GPU of the CUDA of Nvidia company technology also can realize such scheme, and experiment effect is also ideal.
The foregoing is only preferred embodiment of the present invention, in order to limit the present invention, within the spirit and principles in the present invention not all, any modification of making, be equal to replacement, improvement etc., within all should being included in the scope of protection of the invention.
Claims (8)
1. a method that promotes the intellectual analysis performance, is characterized in that, the method comprises the steps:
The video code flow that A, the GPU that supports quick hard decoder and hard-disk cartridge are first resolution to the compressed encoded video code stream hard decoder of input, and the video code flow that is second resolution by the video code flow compressed encoding of this first resolution, the video code flow of the second resolution after described coding is stored in the video memory of described GPU, and this second resolution is less than first resolution;
B, the video code flow of described second resolution is copied to internal memory from described video memory, for CPU, the video code flow in this internal memory is carried out to intellectual analysis.
2. the method for claim 1, is characterized in that, the size of described second resolution is determined by the needed resolution of intellectual analysis.
3. the method for claim 1, is characterized in that, the GPU of the quick hard decoder of described support and hard-disk cartridge is specially: support the GPU of the Quick Sync Video of Intel company technology or the GPU of the support CUDA of Nvidia company technology.
4. the method for claim 1, is characterized in that, described intellectual analysis is virtual line intellectual analysis or the packet loss detection intellectual analysis of stumbling.
5. a device that promotes the intellectual analysis performance, this application of installation is in intellectual analysis equipment, and the device of this lifting intellectual analysis performance has the GPU that supports quick hard decoder and hard-disk cartridge, it is characterized in that, and this device comprises:
The code stream modular converter, for the described GPU of instruction to the compressed encoded video code stream hard decoder of the input video code flow that is first resolution, and the video code flow that this GPU of instruction is second resolution by the video code flow compressed encoding of this first resolution, the video code flow of described second resolution is stored in the video memory of this GPU, and this second resolution is less than first resolution;
The code stream replication module, copy to internal memory for the video code flow by second resolution from described video memory, for CPU, the video code flow in this internal memory carried out to intellectual analysis.
6. device as claimed in claim 5, is characterized in that, the size of described second resolution is determined by the needed resolution of intellectual analysis.
7. device as claimed in claim 5, is characterized in that, the GPU of the quick hard decoder of described support and hard-disk cartridge is specially: support the GPU of the Quick Sync Video of Intel company technology or the GPU of the support CUDA of Nvidia company technology.
8. device as claimed in claim 5, is characterized in that, described intellectual analysis is virtual line intellectual analysis or the packet loss detection intellectual analysis of stumbling.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201310446301.2A CN103491379A (en) | 2013-09-25 | 2013-09-25 | Method and device for improving intelligent analysis performance |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201310446301.2A CN103491379A (en) | 2013-09-25 | 2013-09-25 | Method and device for improving intelligent analysis performance |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN103491379A true CN103491379A (en) | 2014-01-01 |
Family
ID=49831296
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201310446301.2A Pending CN103491379A (en) | 2013-09-25 | 2013-09-25 | Method and device for improving intelligent analysis performance |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN103491379A (en) |
Cited By (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104580837A (en) * | 2015-01-20 | 2015-04-29 | 南京纳加软件有限公司 | Video director engine based on GPU+CPU+IO architecture and using method thereof |
| CN104768063A (en) * | 2015-04-07 | 2015-07-08 | 天脉聚源（北京）教育科技有限公司 | Video coding method and device |
| CN105228000A (en) * | 2015-09-25 | 2016-01-06 | 网宿科技股份有限公司 | A kind of method and system of the complete hardware transcoding based on GPU |
| CN105898315A (en) * | 2015-12-07 | 2016-08-24 | 乐视云计算有限公司 | Video transcoding method and device and system |
| CN106131564A (en) * | 2016-07-28 | 2016-11-16 | 广东西奥物联网科技股份有限公司 | A kind of method and device that camera video stream is processed |
| CN108206937A (en) * | 2016-12-20 | 2018-06-26 | 浙江宇视科技有限公司 | A kind of method and apparatus for promoting intellectual analysis performance |
| CN108924471A (en) * | 2018-09-13 | 2018-11-30 | 福建星网智慧科技股份有限公司 | A kind of conference system based on QSV fast video encoding and decoding |
| CN113099183A (en) * | 2021-04-08 | 2021-07-09 | 天地伟业技术有限公司 | Intelligent video analysis method and device and electronic equipment |
| WO2022141115A1 (en) * | 2020-12-29 | 2022-07-07 | 深圳市大疆创新科技有限公司 | Video processing method and apparatus, system on chip, and storage medium |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101577110A (en) * | 2009-05-31 | 2009-11-11 | 腾讯科技（深圳）有限公司 | Method for playing videos and video player |
| CN101908035A (en) * | 2010-07-30 | 2010-12-08 | 北京华傲精创科技开发有限公司 | Video coding and decoding method, GPU (Graphics Processing Unit) as well as interacting method and system of same and CPU (Central Processing Unit) |
| CN102273205A (en) * | 2008-11-04 | 2011-12-07 | 先进微装置公司 | Software video transcoder with gpu acceleration |
| CN103023872A (en) * | 2012-11-16 | 2013-04-03 | 杭州顺网科技股份有限公司 | Cloud game service platform |
-
2013
- 2013-09-25 CN CN201310446301.2A patent/CN103491379A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102273205A (en) * | 2008-11-04 | 2011-12-07 | 先进微装置公司 | Software video transcoder with gpu acceleration |
| CN101577110A (en) * | 2009-05-31 | 2009-11-11 | 腾讯科技（深圳）有限公司 | Method for playing videos and video player |
| CN101908035A (en) * | 2010-07-30 | 2010-12-08 | 北京华傲精创科技开发有限公司 | Video coding and decoding method, GPU (Graphics Processing Unit) as well as interacting method and system of same and CPU (Central Processing Unit) |
| CN103023872A (en) * | 2012-11-16 | 2013-04-03 | 杭州顺网科技股份有限公司 | Cloud game service platform |
Non-Patent Citations (1)
| Title |
|---|
| E. EKMEKCIOGLU, S.T. WORRALL, A. M. KONDOZ: "Bit-Rate Adaptive Downsampling for the Coding of Multi-View Video with Depth Information", 《 CONFERENCE:THE TRUE VISION-CAPTURE.TRANSMISSION AND DISPLAY OF 3D VIDEO》 * |
Cited By (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104580837A (en) * | 2015-01-20 | 2015-04-29 | 南京纳加软件有限公司 | Video director engine based on GPU+CPU+IO architecture and using method thereof |
| CN104768063A (en) * | 2015-04-07 | 2015-07-08 | 天脉聚源（北京）教育科技有限公司 | Video coding method and device |
| WO2017050116A1 (en) * | 2015-09-25 | 2017-03-30 | 网宿科技股份有限公司 | Gpu-based complete hardware transcoding method and system |
| CN105228000A (en) * | 2015-09-25 | 2016-01-06 | 网宿科技股份有限公司 | A kind of method and system of the complete hardware transcoding based on GPU |
| CN105228000B (en) * | 2015-09-25 | 2018-10-30 | 网宿科技股份有限公司 | A kind of method and system of the complete hardware transcoding based on GPU |
| WO2017096823A1 (en) * | 2015-12-07 | 2017-06-15 | 乐视控股（北京）有限公司 | Method, apparatus and system for video transcoding |
| CN105898315A (en) * | 2015-12-07 | 2016-08-24 | 乐视云计算有限公司 | Video transcoding method and device and system |
| CN106131564A (en) * | 2016-07-28 | 2016-11-16 | 广东西奥物联网科技股份有限公司 | A kind of method and device that camera video stream is processed |
| CN108206937A (en) * | 2016-12-20 | 2018-06-26 | 浙江宇视科技有限公司 | A kind of method and apparatus for promoting intellectual analysis performance |
| CN108206937B (en) * | 2016-12-20 | 2020-05-19 | 浙江宇视科技有限公司 | Method and device for improving intelligent analysis performance |
| CN108924471A (en) * | 2018-09-13 | 2018-11-30 | 福建星网智慧科技股份有限公司 | A kind of conference system based on QSV fast video encoding and decoding |
| WO2022141115A1 (en) * | 2020-12-29 | 2022-07-07 | 深圳市大疆创新科技有限公司 | Video processing method and apparatus, system on chip, and storage medium |
| CN113099183A (en) * | 2021-04-08 | 2021-07-09 | 天地伟业技术有限公司 | Intelligent video analysis method and device and electronic equipment |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN103491379A (en) | Method and device for improving intelligent analysis performance | |
| CN103702124B (en) | A kind of transcoding accelerated method and transcoder | |
| US9906789B2 (en) | Encoding or decoding method and apparatus | |
| CN105163127B (en) | video analysis method and device | |
| CN102447906A (en) | Low-latency video decoding | |
| CN105874800B (en) | Syntax resolver and syntax analytic method | |
| CN104980764A (en) | Parallel coding/decoding method, device and system based on complexity balance | |
| CN103533353B (en) | A kind of near video coding system | |
| CN104575427A (en) | High-compression-ratio lossless image compression method suitable for embedded system decoding | |
| US8737467B2 (en) | Information processing apparatus and method | |
| CN103945221B (en) | Video coding-decoding method, device and the video code flow of generation | |
| CN103634603A (en) | Method, device and system for video encoding and decoding | |
| MY166354A (en) | Video predictive encoding device, video predictive encoding method, video predictive encoding program, video predictive decoding device, video predictive decoding method, and video predictive decoding program | |
| CN103152570B (en) | A kind of video bit stream coding/decoding method and device | |
| TWI735297B (en) | Coding of video and audio with initialization fragments | |
| CN104519360B (en) | A kind of compression method based on HEVC | |
| CN204131646U (en) | A kind of digital video signal decoder | |
| CN209517383U (en) | Distributed cloud transcoding device for the source preview of distributed video processor signal | |
| CN103096087B (en) | A kind of image and video coding-decoding method and system | |
| Kimura et al. | A full HD multistandard video codec for mobile applications | |
| CN203859835U (en) | Device supporting monitoring video rapid decoding | |
| Trojahn et al. | A comparative analysis of media processing component implementations for the Brazilian digital TV middleware | |
| CN105592316A (en) | Digital video signal decoder | |
| Meng et al. | Software Solution for HEVC Encoding and Decoding | |
| Duan et al. | The Implement of MPEG-4 Video Encoding Based on NiosII Embedded Platform |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20140101 |