CN104410799A - Distributed technical review method - Google Patents
Distributed technical review method Download PDFInfo
- Publication number
- CN104410799A CN104410799A CN201410811306.5A CN201410811306A CN104410799A CN 104410799 A CN104410799 A CN 104410799A CN 201410811306 A CN201410811306 A CN 201410811306A CN 104410799 A CN104410799 A CN 104410799A
- Authority
- CN
- China
- Prior art keywords
- skill
- segmentation
- file
- sub
- video
- 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.)
- Granted
Links
Landscapes
- Two-Way Televisions, Distribution Of Moving Picture Or The Like (AREA)
Abstract
The invention relates to a distributed technical review method. The method comprises the following steps: receiving a technical review file; segmenting the file; performing sectional technical review; and merging reports, wherein the step of segmenting the file comprises the following sub-steps: determining the number of segments; searching the longest duration of video technical review; searching the longest duration of audio technical review; converting into duration in frame unit; acquiring the superposition length; determining the segment length; and segmenting the file. According to the method disclosed by the invention, the GPU computing capability of local resources is fully utilized by virtue of a task segmenting mode. A segmented superposition area is set under the condition that the characteristics and needs of the technical review are fully considered, and the technical review quality is not influenced by the segments. Compared with a traditional technical review method, the method disclosed by the invention has the advantages that local computing resources are fully utilized, the computing capability which cannot be utilized in the traditional method is integrated into a system method, the benefits of single server resources in the traditional method are extended to more local resources, and the aim of rapidly and efficiently executing a task is achieved.
Description
Technical field
The present invention relates to a kind of method that distributed skill is examined, the method for electronic digit process, is a kind of method of process of testing to digital video file.
Background technology
Along with the increase of TV station's traffic carrying capacity, the efficiency that biometrics program skill is examined often cannot meet the standby requirement broadcast of program, the particularly technological audit of coded format and video quality, need the hardware resource of at substantial, program making defective returned reform or revise time, program completes amendment and carries out technological audit again, current skill examine efficiency cannot ensure program efficiently by skill examine, thus affect the standby progress broadcast of program, cause program cannot broadcast at official hour.
Summary of the invention
In order to overcome the problem of prior art, the present invention proposes a kind of method that distributed skill is examined.The mode that described method uses distribution skill to examine, examined by the examination burst of more coded format consuming time and video quality, multithreading is examined, and improves the operating efficiency that skill is examined.
The object of the present invention is achieved like this: a kind of method that distributed skill is examined, the system that described method uses comprises: the skill arranged in server or server cluster or Cloud Server examines the skill document examination unit of file reception unit, segmenting unit, the video content detecting and alarm being provided with multithreading and code/decode format detecting and alarm, the report merge cells with technical result queue, and the step of described method is as follows:
Receive the step that skill examines file: examine file reception unit for skill and receive the source file requiring technological audit;
The step of file cutting: carry out cutting to source file for segmenting unit, forms multiple segmental file;
The step that segmentation skill is examined: respectively coded format examination and video quality examination are carried out to each segmental file for skill document examination unit, form the examining report of each segmentation;
The step that report merges: for reporting that the examining report of merge cells to each segmentation merges, form the examining report that whole skill examines file;
Described " step of file cutting " comprises following sub-step:
Determine the sub-step of segmentation number: for according to dicing strategy determination segmentation number, partition strategy comprises and carries out segmentation by file size or examine the idle number determination segmentation number of node according to skill;
Find the sub-step of the longest duration that video skill is examined: examining duration threshold parameter in item for traveling through all video skills, obtaining video skill and examining maximum duration;
Find the sub-step that audio frequency skill examines the longest duration: examining duration threshold parameter in item for traveling through all audio frequency skills, obtaining audio frequency skill and examining maximum duration;
Be converted to the sub-step of the duration of frame unit: for according to video formats, audio frequency duration is in seconds converted to the duration of frame unit;
Obtain the sub-step of coincidence length: examine maximum duration and using the maximum in the audio frequency duration in units of frame as the coincidence length in units of frame for getting video skill;
Determine the sub-step of section length: for according to segmentation number, coincidence length, video file total length, determine the length of each segmentation, its in and out point is all in units of frame;
The sub-step of file fragmentation: skill is examined file fragmentation for the length according to segmentation.
Further, described " step that segmentation skill is examined " comprises following sub-step:
The sub-step of decoding: decode to segmental file for the position according in and out point, is decoded as the video data without compression;
The sub-step that parallel skill is examined: examine node for skill and start different threads and call video content detecting and alarm respectively and code/decode format detecting and alarm detects the video/audio without compression.
Further, described " report combining step " comprises following sub-step:
Store the sub-step of first testing result: for the testing result in first segmentation is added Ji Shen result queue;
The sub-step of next segmentation of sequence detection: for traveling through the first testing result in next segmentation, if in Ji Shen result queue last testing result enter out that an interval comprises this testing result enter out an interval, then continue the next testing result in this section of scanning, until its enter out that an interval do not comprise that skill examines last testing result in queue enter out an interval, form final Ji Shen result queue;
Judge the interval sub-step whether intersected: for judging whether former and later two intervals intersect, if "Yes", final skill is examined last testing result interval in queue be modified to these two also interval; If "No", the testing result that in this testing result and this segmentation, this skills all examine item is joined in final Ji Shen result queue below;
Judge whether the sub-step incorporating all segmentation result: for after all testing results of segmentation are added final Ji Shen result queue, judge whether it is incorporated all segmentation result, if "No", get back to " sub-step of next segmentation of sequence detection ", if "Yes", terminate to merge, output detections is reported.
The beneficial effect that the present invention produces is: the GPU computing capability of local resource is fully utilized by the mode of task segmentation by method of the present invention.When taking into full account the Liquified gas tanker that skill is examined, setting up the overlapping region of segmentation, ensureing that segmentation does not affect skill and examines quality.Compared with examining method with traditional skill, the method of the invention takes full advantage of local computational resource, computing capability unserviceable in conventional method is incorporated in systems approach, the interests of Single-Server resource in conventional method are expanded to local more resource, reach tasks carrying object rapidly and efficiently.
Accompanying drawing explanation
Below in conjunction with drawings and Examples, the invention will be further described.
Fig. 1 is the system schematic used described in method described in embodiments of the invention one;
Fig. 2 is the flow chart of method described in embodiments of the invention one;
Fig. 3 is segment logic schematic diagram in method described in embodiments of the invention one;
Fig. 4 is method described in embodiments of the invention three " report combining step " schematic flow sheet.
Embodiment
Embodiment one:
The present embodiment is a kind of method that distributed skill is examined, the system that described method uses comprises: the skill be arranged in server or server cluster or Cloud Server examines the skill document examination unit of file reception unit, segmenting unit, the video content detecting and alarm being provided with multithreading and code/decode format detecting and alarm, the report merge cells with Ji Shen result queue, as shown in Figure 1.
The system used described in method described in the present embodiment can realize by mode flexibly, can be embodied as and comprise a set of server that multiple skill examines node cluster, also each software module can be deployed in cloud computing platform to realize the automatic skill of cloud and examine.If server can be deployed in the user such as TV station, Network TV Station building environment; If the automatic skill of cloud is examined, then provide skill to examine service by privately owned cloud or publicly-owned cloud mode, can be deployed in the privately owned cloud environment of user, or be deployed on publicly-owned cloud platform.The system that method described in the present embodiment uses possesses very large retractility, and point automatic skill of cloudlet is examined node and not etc., can not made full use of cloud resource, improve the efficiency that skill is careful by a node from several node to hundreds of.
The result that report content=file format detection+code/decode format detection+audio-visual content that complete distributed skill is examined detects.Wherein, file format detects because efficiency very high (two treatment effeciency to strong E5-2600 platform is more than 10 speeds), adds file format and detects the problem also having file redirect, therefore does not carry out distributed skill and examine, and only carries out the detection of whole section.Code/decode format detects and audio-visual content detects execution segment processing.
The concrete steps of the method described in the present embodiment are as follows, flow chart as shown in Figure 2:
Receive the step that skill examines file: examine file reception unit for skill and receive the source file requiring technological audit.Source file is sent by the system of asking skill to be examined, and the system that these request skills are examined can be the system of store video files or the system of making video file, or the various video files transmitted by satellite or network etc.
The step of file cutting: carry out cutting to source file for segmenting unit, forms multiple segmental file.File cutting must take into full account and prevent the feature that skill is examined undetected.
The segmentation principle of file cutting examines parameter according to skill to carry out the most rational segmentation, neither there is redundancy segmentation, also can not produce undetected, with ensure the high efficiency that parallel skill is examined and skill conclude fruit correctness.File cutting will carry out cutting according to the requirement that video content detects and code/decode format detects two aspects.First video content detecting and alarm knows the number of section according to set dicing strategy, and dicing strategy comprises and undertaken cutting into slices by file designated length or determine section according to the idle number that skill examines node.
Because a lot of detection is that front and back are relevant, such as quiet detection wants quiet perdurabgility to exceed the just meeting alarm of certain threshold values, need to consider that overlay region is to ensure that burst logical process is accurate, video content detecting and alarm can calculate the in and out point information of each segmentation.
The length requirement of overlay region mainly considers that video skill examines the duration threshold parameter in project, these duration threshold values, comprising: the time threshold values that the folder frame that the uncorrelated frame within the time threshold values that the quiet frame that time threshold values, continuous how many frames that the black field continuous how many frames are detected as black field is detected are detected as quiet frame detects, continuous how many frames can be regarded as folder frame detects, continuous how many frames are detected as the exceed standard time threshold values detected, continuous how many frames of YUV that YUV exceeds standard is detected as brightness that brightness exceeds standard and exceeds standard the time threshold values etc. detected.Also have simultaneously consider audio frequency skill examine in duration threshold value, comprising: how long be detected as the time threshold values of quiet quiet detection continuously, how long be detected as the time threshold values of the anti-phase detection of the anti-phase audio frequency of audio frequency continuously, how long be detected as continuously real peak value that real peak value exceeds standard exceed standard detect time threshold values, how long be detected as loudness that loudness exceeds standard continuously and exceed standard the time threshold values etc. detected.Because audio frequency duration threshold value is all using second as unit, and video duration threshold value is all using frame as unit, in order to the skill looking after video is examined, therefore, also will become using second as the unit conversion of duration using frame as duration unit, to contrast.Again the longest video duration threshold parameter and the longest audio frequency duration threshold parameter are contrasted, get the length of its this effect overlay region maximum.
" step of file cutting " comprises following sub-step:
Determine the sub-step of segmentation number: for according to dicing strategy determination segmentation number, partition strategy comprises and carries out segmentation by file size or examine the idle number determination segmentation number of node according to skill.
Find the sub-step of the longest duration that video skill is examined: examining duration threshold parameter in item for traveling through all video skills, obtaining video skill and examining maximum duration.
Find the sub-step that audio frequency skill examines the longest duration: examining duration threshold parameter in item for traveling through all audio frequency skills, obtaining audio frequency skill and examining maximum duration.
Be converted to the sub-step of the duration of frame unit: for according to video formats, audio frequency duration is in seconds converted to the duration of frame unit.
Obtain the sub-step of coincidence length: examine maximum duration and using the maximum in the audio frequency duration in units of frame as the coincidence length in units of frame for getting video skill.
Determine the sub-step of section length: for according to segmentation number, coincidence length, video file total length, determine the length of each segmentation, its in and out point is all in units of frame.
The sub-step of file fragmentation: skill is examined file fragmentation for the length according to segmentation.The content of segmentation comprises two parts: code/decode format detects (segmental file encoding and decoding detection)+audio-visual content and detects (detection of segmental file audio-visual content).
The segmentation one side needs that audio-visual content detects are known that according to partition strategy the number of section, partition strategy comprise and are carried out segmentation or the idle number determination segments according to segmented node by file designated length.Need on the other hand to consider because a lot of detection is that front and back are relevant, such as quiet detection wants quiet perdurabgility to exceed the just meeting alarm of certain threshold values, bad will the causing of segment logic process is failed to judge, and need the segment logic shown in design drawing 3, emphasis will have overlay region exactly for this reason.Method is the duration parameters that all skills of statistics examine item, gets the minimum coincidence length of maximum duration as segmentation, completes segmentation divide (the several step process in the face that sees above) according to segmentation parameter and coincidence length.
The segmentation that code/decode format detects, on the basis needing the segmentation detected at audio-visual content to complete, will align to MPEG2 IBP GOP.
The step that segmentation skill is examined: respectively coded format examination and video quality examination are carried out to each segmental file for skill document examination unit, form the examining report of each segmentation.Fragmentation technique first skill examines node according to the position entered out a little to detected file decoding, is decoded as the video/audio without compression.Skill is examined node and is started different threads and call video content detecting and alarm and code/decode format detecting and alarm respectively and detect the video/audio without compression.
The step that report merges: for reporting that the examining report of merge cells to each segmentation merges, form the examining report that whole skill examines file.During merging, need to carry out duplicate removal and merging treatment to each testing result in order.Utilize multithreading, detection is examined to all skills, travel through the testing result of every section in order, form examining report.
Distributed skill examine process that splicing merges be exactly in order to each testing result carry out duplicate removal the process of merging treatment.Get the testing result of each detection segment, and add up.Report duplicate removal afterwards, final report (FR) merges.
In addition, when distributed skill is examined, frame detects can not reprocessing for overlay region, can further improve efficiency, decreases duplicate removal work when frame testing result merges.
Distributed skill described in the present embodiment is examined can realize the parallel processing of many computing nodes, according to computing node number, substantially can reach linear increment.The automatic skill that the automatic skill not performing distributed treatment is examined task and performed distributed treatment examines automatic skill that task obtains, and to examine report completely the same, and ensure the objective and accurate of data, the exchange being also conducive to data message is shared and statistical analysis.
Embodiment two:
Originally being the improvement of embodiment one, is the refinement of embodiment one about " step that segmentation skill is examined "." step that segmentation skill is examined " described in the present embodiment comprises following sub-step:
The sub-step of decoding: decode to segmental file for the position according in and out point, is decoded as the video data without compression.Due to file segmentation, this step can utilize multiple thread to decode to each segmentation simultaneously, accelerates decoding speed.
The sub-step that parallel skill is examined: examine node for skill and start different threads and call video content detecting and alarm respectively and code/decode format detecting and alarm detects the video/audio without compression.Skill examines each the frame Audio and Video data by analyzing detected material, and the data parsed and relevant criterion are carried out calculating contrast, thus determine whether containing various video and audio problem in tested material, and thinking problematic local marking point record, for manually rechecking.
Skill is examined content and is comprised: audio-visual content detects and code/decode format detects.Video content is detected and can call the main contents that multiple thread detects simultaneously and comprise: black field, colour bar, the blue end, the green end, quiet frame, station symbol, Y/U/V exceeds standard, R/G/B exceeds standard, width type, flower screen detect.Audio frequency aspect mainly comprises: quiet, distorsion, VU, level, stereo phase place, loudness such as to exceed standard at the detection.In addition, examine the detection range of parameter by adjustment skill, the strategy that adjustment detects, accurately can locate that program file exists look, audio issue.
Embodiment three:
The present embodiment is the improvement of above-described embodiment, above-described embodiment about the refinement of " report combining step ", as shown in Figure 4." report combining step " described in the present embodiment comprises following sub-step:
Store the sub-step of first testing result: for the testing result in first segmentation is added Ji Shen result queue.In the data meeting dynamic change of the process Zhong Jishen result queue merged, until become final Ji Shen result queue after having merged.
The sub-step of next segmentation of sequence detection: for traveling through the first testing result in next segmentation, if in Ji Shen result queue last testing result enter out that an interval comprises this testing result enter out an interval, then continue the next testing result in this section of scanning, until its enter out that an interval do not comprise that skill examines last testing result in queue enter out an interval, form final Ji Shen result queue.This sub-step is report duplicate removal, duplicate removal principle: the skill that front and back are two sections concludes entering out an interval and whether there is inclusion relation of fruit, if existed, then deletes subinterval.
Judge the interval sub-step whether intersected: for judging whether former and later two intervals intersect, if "Yes", final skill to be examined in queue last testing result interval be modified to these two also interval, if "No", the testing result that in this testing result and this segmentation, this skills all examine item is below joined in final Ji Shen result queue.This sub-step merges for report, combination principle: the skill that front and back are two sections concludes entering out an interval and whether existing to intersect and do not comprise key of fruit, if existed, then interval to merge two.
Judge whether the sub-step incorporating all segmentation result: for after all testing results of segmentation are added final Ji Shen result queue, judge whether it is incorporated all segmentation result, if "No", get back to " sub-step of next segmentation of sequence detection ", if "Yes", terminate to merge, output detections is reported.
File audio-visual content testing result complete after file Codec testing result+merging that complete skill is complete after examining report content=file format test result+merging.Skill examine report complete after in write into Databasce, feed back to the initiator that request skill is examined simultaneously.
Finally it should be noted that, below only in order to technical scheme of the present invention to be described and unrestricted, although with reference to preferred arrangement scheme to invention has been detailed description, those of ordinary skill in the art is to be understood that, can modify to technical scheme of the present invention (sequencing etc. of the calculating of such as section length, the calculating of overlap length, step) or equivalent replacement, and not depart from the spirit and scope of technical solution of the present invention.
Claims (3)
1. the method that a distributed skill is careful, the system that described method uses comprises: the skill be arranged in server or server cluster or Cloud Server examines the skill document examination unit of file reception unit, segmenting unit, the video content detecting and alarm being provided with multithreading and code/decode format detecting and alarm, the report merge cells with technical result queue, and the step of described method is as follows:
Receive the step that skill examines file: examine file reception unit for skill and receive the source file requiring technological audit;
The step of file cutting: carry out cutting to source file for segmenting unit, forms multiple segmental file;
The step that segmentation skill is examined: respectively coded format examination and video quality examination are carried out to each segmental file for skill document examination unit, form the examining report of each segmentation;
The step that report merges: for reporting that the examining report of merge cells to each segmentation merges, form the examining report that whole skill examines file;
It is characterized in that, described " step of file cutting " comprises following sub-step:
Determine the sub-step of segmentation number: for according to dicing strategy determination segmentation number, partition strategy comprises and carries out segmentation by file size or examine the idle number determination segmentation number of node according to skill;
Find the sub-step of the longest duration that video skill is examined: examining duration threshold parameter in item for traveling through all video skills, obtaining video skill and examining maximum duration;
Find the sub-step that audio frequency skill examines the longest duration: examining duration threshold parameter in item for traveling through all audio frequency skills, obtaining audio frequency skill and examining maximum duration;
Be converted to the sub-step of the duration of frame unit: for according to video formats, audio frequency duration is in seconds converted to the duration of frame unit;
Obtain the sub-step of coincidence length: examine maximum duration and using the maximum in the audio frequency duration in units of frame as the coincidence length in units of frame for getting video skill;
Determine the sub-step of section length: for according to segmentation number, coincidence length, video file total length, determine the length of each segmentation, its in and out point is all in units of frame;
The sub-step of file fragmentation: skill is examined file fragmentation for the length according to segmentation.
2. method according to claim 1, is characterized in that, described " step that segmentation skill is examined " comprises following sub-step:
The sub-step of decoding: decode to segmental file for the position according in and out point, is decoded as the video data without compression;
The sub-step that parallel skill is examined: examine node for skill and start different threads and call video content detecting and alarm respectively and code/decode format detecting and alarm detects the video/audio without compression.
3. method according to claim 1 and 2, is characterized in that, described " report combining step " comprises following sub-step:
Store the sub-step of first testing result: for the testing result in first segmentation is added Ji Shen result queue;
The sub-step of next segmentation of sequence detection: for traveling through the first testing result in next segmentation, if in Ji Shen result queue last testing result enter out that an interval comprises this testing result enter out an interval, then continue the next testing result in this section of scanning, until its enter out that an interval do not comprise that skill examines last testing result in queue enter out an interval, form final Ji Shen result queue;
Judge the interval sub-step whether intersected: for judging whether former and later two intervals intersect, if "Yes", final skill is examined last testing result interval in queue be modified to these two also interval; If "No", the testing result that in this testing result and this segmentation, this skills all examine item is joined in final Ji Shen result queue below;
Judge whether the sub-step incorporating all segmentation result: for after all testing results of segmentation are added final Ji Shen result queue, judge whether it is incorporated all segmentation result, if "No", get back to " sub-step of next segmentation of sequence detection ", if "Yes", terminate to merge, output detections is reported.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410811306.5A CN104410799B (en) | 2014-12-24 | 2014-12-24 | A kind of careful method of distributed skill |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410811306.5A CN104410799B (en) | 2014-12-24 | 2014-12-24 | A kind of careful method of distributed skill |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN104410799A true CN104410799A (en) | 2015-03-11 |
| CN104410799B CN104410799B (en) | 2017-10-31 |
Family
ID=52648386
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201410811306.5A Active CN104410799B (en) | 2014-12-24 | 2014-12-24 | A kind of careful method of distributed skill |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104410799B (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104837033A (en) * | 2015-05-20 | 2015-08-12 | 腾讯科技（北京）有限公司 | Information processing method and server |
| CN105488100A (en) * | 2015-11-18 | 2016-04-13 | 国信司南(北京)地理信息技术有限公司 | Efficient detection and discovery system for secret-associated geographic data in non secret-associated environment |
| CN105578203A (en) * | 2016-01-15 | 2016-05-11 | 北京赛特斯信息科技股份有限公司 | Video file quality detection system and method |
| CN113225623A (en) * | 2021-04-30 | 2021-08-06 | 四川云从天府人工智能科技有限公司 | Adaptive video file accelerated parsing method, computer device and storage medium |
| CN116489421A (en) * | 2023-05-08 | 2023-07-25 | 湖南芒果无际科技有限公司 | Video AI intelligent detection method and system based on file slicing |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102006499A (en) * | 2010-12-10 | 2011-04-06 | 北京中科大洋科技发展股份有限公司 | Method for detecting video and audio quality of digital television program file |
| US20110271307A1 (en) * | 2009-12-18 | 2011-11-03 | Tektronix International Sales Gmbh | Video data stream evaluation systems and methods |
| CN103873852A (en) * | 2012-12-11 | 2014-06-18 | 上海文广互动电视有限公司 | Multi-mode parallel video quality fault detection method and device |
-
2014
- 2014-12-24 CN CN201410811306.5A patent/CN104410799B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20110271307A1 (en) * | 2009-12-18 | 2011-11-03 | Tektronix International Sales Gmbh | Video data stream evaluation systems and methods |
| CN102006499A (en) * | 2010-12-10 | 2011-04-06 | 北京中科大洋科技发展股份有限公司 | Method for detecting video and audio quality of digital television program file |
| CN103873852A (en) * | 2012-12-11 | 2014-06-18 | 上海文广互动电视有限公司 | Multi-mode parallel video quality fault detection method and device |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104837033A (en) * | 2015-05-20 | 2015-08-12 | 腾讯科技（北京）有限公司 | Information processing method and server |
| CN104837033B (en) * | 2015-05-20 | 2018-09-25 | 腾讯科技（北京）有限公司 | A kind of information processing method and server |
| CN105488100A (en) * | 2015-11-18 | 2016-04-13 | 国信司南(北京)地理信息技术有限公司 | Efficient detection and discovery system for secret-associated geographic data in non secret-associated environment |
| CN105578203A (en) * | 2016-01-15 | 2016-05-11 | 北京赛特斯信息科技股份有限公司 | Video file quality detection system and method |
| CN105578203B (en) * | 2016-01-15 | 2018-10-19 | 北京赛特斯信息科技股份有限公司 | Video file quality detecting system and method |
| CN113225623A (en) * | 2021-04-30 | 2021-08-06 | 四川云从天府人工智能科技有限公司 | Adaptive video file accelerated parsing method, computer device and storage medium |
| CN116489421A (en) * | 2023-05-08 | 2023-07-25 | 湖南芒果无际科技有限公司 | Video AI intelligent detection method and system based on file slicing |
Also Published As
| Publication number | Publication date |
|---|---|
| CN104410799B (en) | 2017-10-31 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN104410799A (en) | Distributed technical review method | |
| CN103873852A (en) | Multi-mode parallel video quality fault detection method and device | |
| CN102006499B (en) | Method for detecting video and audio quality of digital television program file | |
| CN111641869B (en) | Video split mirror method, video split mirror device, electronic equipment and computer readable storage medium | |
| US20230094834A1 (en) | Fault Detection Method and Device of Live Stream Service, Electronic Device, and Readable Storage Medium | |
| CN105491395B (en) | Server video management method and system | |
| CN103176996A (en) | Image search method based on image feature information and image search engine server based on image feature information | |
| CN112055198A (en) | Video testing method and device, electronic equipment and storage medium | |
| EP4071616A1 (en) | Method for generating topology diagram, anomaly detection method, device, apparatus, and storage medium | |
| CN108134746B (en) | Method and device for processing rail transit data | |
| WO2018177018A1 (en) | Multi-tuner set-top box production detection method, system and detection device | |
| CN109040784A (en) | Commercial detection method and device | |
| CN113691803A (en) | Method, device, equipment and medium for testing audio and video interface function | |
| CN112788356A (en) | Live broadcast auditing method, device, server and storage medium | |
| CN106777010B (en) | Log providing method and device and log obtaining method, device and system | |
| CN111737371A (en) | Data flow detection classification method and device capable of dynamically predicting | |
| CN109325166B (en) | Method and device for configuring analysis rules in crawler system | |
| CN113015198B (en) | Method, device and equipment for positioning fault base station and readable storage medium | |
| CN111061712A (en) | Data connection operation processing method and device | |
| CN112804070A (en) | Method, device and equipment for positioning service barrier | |
| CN115953715B (en) | Video detection method, device, equipment and storage medium | |
| CN112328838B (en) | Message analysis preprocessing method, system and storage medium | |
| CN117278439B (en) | Communication quick verification method based on compression algorithm | |
| CN104298692B (en) | A kind of method and system of distributed search | |
| US20230269142A1 (en) | Network data analysis method, network data analysis functional network element and communication system |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |