IQanalytics User Manual
Contents
1. Object Entered It is also possible to store event images and videos upon receiving events from IVA To enable this feature please contact IQinVision for license details for the events module 49 48 47 46 45 44 43 42 40 39 37 36 35 34 33 32 J 39 37 5 EE 31 0 5 Camera Placement amp Guidelines This section provides recommended camera angles and positions Camera Positioning Camera positioning is critical for capturing good data for video analytics Care should be taken to avoid a object size being set too large and b occlusions Typically cameras should be installed with one of three views 1 Top Down view vertical ceiling mount 2 Perspective or Angled view wall or corner mount looking down Useful for detecting general intrusion 3 Side View horizontal wall mount at lower height The Top Down View Is generally good for counting detecting direction of movement etc The second option perspective or angle view is good for detecting intrusions The third option side view should only be used for face detection It is not recommended for other analytics since objects can be hidden from view occlusions FOV Field of View and Object Size For people detection the object or person size should be 10 20 of the whole scene The object size should not be set to occupy 40 100 of the scene Recommended Scene and Lighting Conditions General recommendations for lighting w2. ignore area these zones will be ignored during processing 9 Ensure that the size of the objects that are to be tracked in the scene are not too small Object sizes have to be at least 5 10 of the scene for consistent detection If the objects are only a few pixels people or vehicles it will be difficult to detect them consistently and will cause false alarms on most systems 10 Ensure that the object size is not too large more than 40 of the scene The camera should not be too close to the objects being tracked 11 We recommend you turn off AGC Auto Gain Auto IRIS and White Balance on cameras These fluctuate some times and cause notable scene changes It is possible to run the system with these settings ON with lower sensitivity if the scene Is Stable and fluctuations are infrequent 12 Camera lenses need to be kept clean and focused for best results Camera Zoom FOV and Angle are important The following sections address them fully with examples Recommended Camera Positioning for Common Applications 1 Intrusion Detector Camera must be at an angle looking down Camera should be installed at sufficient height to avoid occlusions The object size should not be too large relative to the scene It is recommended that the size of any object entering the scene not exceed 50 of the scene size If the camera is used during the night lighting reflections should be minimized to avoid false detections Light should be
3. high it is also advisable to turn on the Continuous Learning feature to decrease false detections This results in objects being detected properly in low light conditions 7 Event Output Format When an event is generated IQanalytics will provide event metadata and object metadata associated with that event This metadata will be included in the JPEG headers or H264 Headers Please contact IQeye on details for parsing this data from the stream The event information is provided in the following xml structure Aside from the camera tamper event every other event will be accompanied by object data that gives the ID Location and Size of the object triggering the event lt EventTypexEVENT_INTRUSION lt EventType gt lt ObjectData gt lt gt lt l ObjectID 3 gt lt LeftSide gt 20 lt LeftSide gt lt TopSide 40 lt TopSide gt lt RightSide 50 lt RightSide lt BottomSide gt 80 lt BottomSide gt lt Areard27 lt Area gt lt gt lt ObjectData gt a The Event Type can contain following values EVENT_INTRUSION EVENT TAMPER EVENT ENTER Motion Detection EVENT_EXIT b The Object ID is the ID of the object that created the event One object can create multiple events For motion detection an object will create an Entry and Exit event to signal when the object entered and left the camera FOV c The Object Location values in in the ObjectData are mean positions of the object rectangle sides In pi
4. object sizes and sensitivity For example Modified Settings Image Two objects soot A typical rule of detected as thumb when setting one gt minimum and 1 maximum sizes is to set the minimum size of an object to half the width and height a of the average object Pea In this scene two objects are and the maximum We can see that adjusting the being detected as one This is size of an object to maximum size allows the two persons usually due to the maximum about 130 the to be detected as separate objects size of objects not set correctly width and height of In this situation reduce the the average object One object detected as two In this scene one object is detected as two separate objects This is usually due to the minimum and or maximum size set incorrectly In this situation increase the minimum and or maximum size of objects accordingly to accommodate larger objects In some scenes the video contains a high amount of noise that greatly affects the engine by causing unwanted objects to be detected and or trigger unwanted events A typical rule of thumb when setting minimum and maximum sizes is to set the minimum size of an object to half the width and height of the average object and the maximum size of an object to about 130 the width and height of the average object To minimize the effects of noise Sensitivity can be reduced so that only more prominent objects will be We can
5. see that increasing the minimum and maximum size allows the object to be correctly detected as one The result from these settings allows the engine to ignore video noise and still be able to detect objects in the detected and trigger events Adjusting Advanced Settings Some scenes require extra or advanced settings like continuous learning shadow Suppression etc Continuous Learning This feature allows the engine to learn the scene as It changes gradually Learning happens in every frame and gradual changes In background are captured This feature Is typically used for scenes such as 1 Qutdoor scenes with shadows 2 Indoor scenes with changing lighting conditions 3 Parking areas There are other features involved in continuous learning such as learn object learning time learn scene etc For example Different Environment Conditions Indoor scene with changing lighting condition Default Settings Image Recommended Settings Modified Settings Result In order to accommodate i a sudden changes in the scene due to lighting conditions Scene Change Based Learning can be turned on Sensitivity can also be reduced to reduce false alarms due to lighting changes In this scene lighting condition changes frequently due to lights being tumed off and on by a sensor With continuous learning on changing lighting conditions can be accommodated so as not If Camera Tamper feature is trigger false al
6. specific settings Tamper Provides Camera Tamper specific settings Advanced Provides Advanced and learning settings for video analytics ey gt Select the IQanalytics app from the applications tab to access the configuration pages IQ765N IQEYE331A60 v ive tup se networ secur event Applications IQanalytics General Settings motion events enabled sensitivity 1 100 B0 Object Size Minimum Width 5 Minimum Height 25 Maximum Width B0 Maximum Height feo IQinVision amp General Settings The following parameters are available for general setting tuning A Motion Events Enable to detect all object enter and object exit events B Sensitivity A higher sensitivity value will detect more motion movements C Object Sizes The Object classification table specifies the allowable maximum width height and minimum width height parameters applicable for different types of objects Setting these values can help fine tune the engine performance By clicking the check box for different object types you can enable disable object detection based on size thresholds Object By clicking the check box you can enable disable the object detection e Object Minimum width It specifies the minimum allowable width of the object to be detected It is in terms of percentage e Object Minimum height specifies the minimum allowable height of the object to be detected in terms of percent
7. IQ inVision IQanalytics Vandal Motion and Intrusion Detection User Guide applications Applications IQanalytics Action 7 status running QED intrusion tamper 9 advanced i General Settings motion events enabled Y sensitivity 1 100 fo Object Size Minimum Width f5 Minimum Height pe Maximum Width 30 Maximum Height 0 IQinVision Contents Overview Installation IQanalytics Configuration Advanced Pages Camera Placement amp Guidelines Recommended Settings Event Output Format NO o1R OD i eR 0 CO O1 no 1 Overview IQanalytics software provides an advanced solution that performs intelligent video analysis and fully automates video monitoring It automatically tracks and identifies objects analyzes motion and extracts video intelligence from the camera These analytics can output real time events and object data for video data mining or storage in a database IQanalytics is focused on automating video analysis and security alerts thus eliminating the need for manual work and huge monitoring costs It also increases productivity and efficiency of video surveillance systems and the people who monitor them The analytics offered are e Motion Detection e Intrusion Detection e Camera Tamper Detection Motion Detection This analytic application analyzes video in real time and detects valid motion ina scene Intrusion Detection Intrusion detection provi
8. age e Object Maximum width specifies the maximum allowable width of the object to be detected in terms of percentage e Object Maximum height specifies the maximum allowable height of the object to be detected in terms of percentage Settings for people vehicle and other type of objects are similarly setup Please note that if more than one type of object is enabled the entire object with sizes within the minimum and maximum set values of height width will get detected Also if object type none is enabled all objects regardless of its height width will be detected Intrusion Settings This page contains Intrusion Detector settings general intrusion tamper advanced Intrusion Settings intrusion enabled brush 3x3 A Enable will enable the Intrusion detection B Brush This allows the selection of a brush size for area selection painting To select an area for Intrusion Detector click and drag on the image to paint the scene The area highlighted in red will be the intrusion area Click and drag on the red area to de select that section of the grid Camera Tamper Settings This notifies you when a camera has been tampered with It looks for blocking de focusing power video loss light changes and drastic changes in camera angle intrusion Camera Tamper Settings camera tamper detector enabled v A Enable This will enable Camera Tamper Detection Ignore Area Settings Enabl
9. arms the scene will cause the Scene Change event to trigger Scene with moving trees Scene with cars parking for extended period of time This outdoor scene includes a lot of unwanted movement These movements can cause the engine to detect them as objects and trigger unwanted events This scene shows a car arriving at a parking lot and parks for a long period of time By default the car would get absorbed learned into the background after a set amount of time In some situations these objects need to be tracked even though they are not moving This scene is set at night time Objects are not as prominent at low light situations and sometimes the engine might not be able to detect them as intended In order to avoid detection of _ tree movement Sensitivity 1 events Another solution is to can be reduced to avoid the trees from triggering any set the area around the trees as Ignored Area These settings will the engine to avoid detecting the movement from the trees To accommodate objects arriving at the scene and staying for long periods of time the Object Learning feature can be tumed on The result of these settings would allow the engine to track the cars that have parked for a long penod of time without them being leamed into the background In scenes with low contrast and low light sensitivity needs to be increased While sensitivity is set to
10. cs Installed in the Camera Application section on security page should be displayed 9 After installation if application is not licensed please contact IQeye for license keys IQ765N IQEYE331A60 ve sion ve seen applications ase Passwords general protection disabled v Change general password Change management password fs retype management password fr Access Hours weekdays on weekends on camera is on 24 hour HH MM camera is off Camera Firmware Information Firmware File Choose File No file chosen firmware 1Q765N Version V3 6 6 140211 expecting igeye76 sys file name iqeye76 sys bootcode 1 19 Camera License On Camera Software Licenses License File Choose File No file chosen IQaccess Installed expecting codeauth 00501a331a60 upload DTS Installed tex 3 c events Installed ores 3 Applicat event video Not Installed ares A n IQrecorder Not Installed Application File Choose File No file chosen Camera Applications expecting application iqa IQanalytics Installed IQinVision amp 3 IQanalytics Configuration The configuration of IQanalytics parameters can be adjusted using the web user interface The IQanalytics configuration has the following sections detailed descriptions are provided in subsequent sections General Provides general settings for motion detection and video analytics Intrusion Provides Intrusion
11. des automated perimeter monitoring and secure area protection It continuously monitors user selected zones or areas for people or vehicle intrusion This analytic issues real time alarms upon detecting true people or vehicle intrusions into a secure area Camera Tamper Detector Camera Tamper Detector monitors the video and automatically detects camera failure or sabotage It detects any attempts to tamper with the camera partially or completely blocking its lens or field of view or drastically changing the camera angle 2 Installation If the IQanalytics Application package is not pre installed on your camera manual installation will be required Before installation please ensure the IQanalytics iga application file is available 1 Connect IQeye camera onto the local network 2 To help detect camera on local network use IQeye Auto run CD that is provided with the camera 3 Click Scan for Qeye cameras on my network to find IP address of the camera 4 Type camera IP address into your browser 5 Goto setup tab in camera web UI if prompted for authentication information the default username is root default password is system 6 Goto security tab in camera web UI 7 Click on the application package file in Application File select IQanalytics iga and press upload Please don t switch off or reset camera until uploading and rebooting is complete 8 After uploading the message IQanalyti
12. hen using video analytics 1 The scene should be well lit Low lighting levels produce video noise that can triggers false object alerts 2 Stable and consistent lighting will provide good results Fluctuations in lighting or uneven lighting at the scene will cause false alarms In any video analytics system 3 If providing appropriate lighting is an issue it is recommended that a camera with IR illumination be used 4 Avoid occlusions one object going in front of another This will break object tracks and will cause object disappearance and other related problems 9 Avoid pointing the camera directly at light sources as described below The following situations will make the video unusable for analytics and will temporarily signal saturation a Camera pointed directly at the sun b Camera pointed directly at light source c Camera looking at vehicles with headlights cars trains in tunnels 6 Avoid or minimize shadows by repositioning the camera or light source s 7 f possible avoid the presence of trees and other vegetation in the camera s field of view Swaying or movement of trees and bushes in the wind can cause false alarms If they cannot be avoided use ignore area these zones will be ignored during processing 8 If possible avoid the presence of water areas rivers lakes etc in the camera s field of view Water areas can cause reflections and ripples that can cause false alarms If they cannot be avoided use
13. ing false detections due to scene or environmental changes D Object Learning Enable this feature to learn stationary objects into the background E Object Learn Time The time in seconds a stationary object is in the scene before IQanalytics will learn the object into the background F Sudden Change Learning Enable this feature to automatically relearn the background if the entire scene changes by more than the specified threshold G Sudden Change Sensitivity The percentage of the scene that needs to change before the sudden change learning Is triggered H Default Settings Resets all parameters back to default settings Relearn Manually relearn the background 4 Advanced Pages A few hidden pages are also provided for testing and demo purposes These pages are not exposed due to its CPU usage but are useful for testing and demo It is not recommended to use these pages during production or live deployment Events Page This page displays a list of the most Event ID Event Description recent events To access this page input Object Exited the following into your web browser Object Entered http lt ipaddress gt 88 event cgi e o oe Events that are supported e Object Entered iVMD e Object Exited iVMD e Intrusion Intrusion Detector _ e Camera Tampered Camera Tamper a ae Detector Object Entered Object Exited Object Entered 41 Object Exited Object Entered Object Exited
14. ing the ignore area allows the engine to ignore processing in areas that don t require it This may include the ceiling moving light sources environmental conditions or areas where no activity is expected This allows processing to be reduced and lowers false alarms general intrusion tamper Ignore Area Settings ignore area disabled Y brush 1x1 M Advanced Settings processing quality Normal v Learning Settings GRR BRGe object learning enabled v object learn time sec feo sudden change learning enabled 7 sudden change sensitivity fo 1 100 A Enable will enable the ignore area B Brush allows area selection painting to use a larger 3x3 area To select an area to ignore click and drag on the image to paint the scene The area highlighted in red will be the ignored Click and drag on the red area to de select that section of the grid Advanced Settings C Processing Quality Select the Processing Quality type from the drop down box depending upon scene The higher quality allows for detecting smaller objects however overall accuracy may be reduced as frame rate will be decreased due to additional processing power requirements The choices are i Low Default Recommended ii Medium iii High iv Original Learning Settings The learning settings effect how IQanalytics learns the background over time Fine tuning these parameters will help in reduc
15. uniformly distributed in the scene avoid shadows In addition ensure no light sources are directed at the camera A camera that is pointed at an angle looking down is optimal for intrusion detection as seen below Angled View Top Down View 2 Camera Tamper Detector For tamper detection the camera can be installed in any position top down view Side view or front view The maximum size of objects entering the scene should not exceed 50 60 of scene size If the camera is used during the night lighting reflections should be minimized to avoid false detections Light should be uniformly distributed in the scene avoid shadows In addition ensure no light sources are directed at the camera S a Angled View Top Down View 6 Recommended Settings Overview The performance of the video analytics engine relies greatly on scene type and the settings applied to those scenes This section provides a collection of common scenes and the corresponding recommended settings for each These recommended settings are meant to be used only as a guide toward obtaining more accurate results Since every scene is different the recommended settings do not show specific values They are simply a guide to what settings need to be adjusted for in different scenarios to improve on certain aspects of the results Adjusting Basic Settings For some scenes the default settings should be sufficient without changing anything other than minimum and maximum
16. xels of the processing picture d The Object Size is provided as the area value of an object in pixel count IQ inVision 901 0360 RevA 33122 Valle Road San Juan Capistrano CA 92675 4853 USA phone 1 949 369 8100 fax 1 949 369 8105 www igeye com
Download Pdf Manuals
Related Search