Difference between revisions of "Retail Analytics/Run"
From DS Help
Paulo Santos (Talk | contribs) |
Paulo Santos (Talk | contribs) |
||
| (19 intermediate revisions by the same user not shown) | |||
| Line 8: | Line 8: | ||
In the left side, you have "Video Source". Select the camera you want to use with the system. Click the play button on the right. | In the left side, you have "Video Source". Select the camera you want to use with the system. Click the play button on the right. | ||
Below this, there's "Video Settings". This is where you can configure some properties of the video source, like brightness or contrast. | Below this, there's "Video Settings". This is where you can configure some properties of the video source, like brightness or contrast. | ||
| − | In the right top corner, you can select "Full Screen" to run the system in full screen. | + | In the right top corner, you can select "Full Screen" to run the system in full screen. To configure your triggers please refer to [[Retail Analytics/Statistics#Triggers|Triggers]]. |
| + | |||
| + | [[File:VideoSettings_RA.png|thumb|250px|none|Video Settings]] | ||
==Advanced Settings== | ==Advanced Settings== | ||
| Line 16: | Line 18: | ||
;Kiosk | ;Kiosk | ||
| − | : In Kiosk mode, the trigger is activated as soon as a person stands in front of the camera. The system | + | : In Kiosk mode, the trigger is activated as soon as a person stands in front of the camera. The system analyses the person's data and starts its trigger, for the respective kind of consumer (by gender, age group, ...). |
;Big Space | ;Big Space | ||
| − | : "Engine Settings" has an option named "Big space tick frequency". This option is defined in minutes and refers to the time interval of analysis. If set to 30 (minutes) this means that, for 30 minutes the engine will | + | : "Engine Settings" has an option named "Big space tick frequency". This option is defined in minutes and refers to the time interval of analysis. If set to 30 (minutes) this means that, for 30 minutes the engine will analyse the consumer's data (gender, age group) without any trigger during that time. After 30 minutes and after find the predominant kind of consumer, the specific trigger will start. |
==Engine Settings== | ==Engine Settings== | ||
| + | |||
| + | [[File:EngineSettings_RA.png|thumb|250px|none|Engine Settings]] | ||
;Learn time | ;Learn time | ||
| − | : | + | : After recognize a face, this is the amount of time spent analyzing data (gender, age, ...) that is stored in Database. |
;Minimum face size | ;Minimum face size | ||
| − | : | + | : Minimum needed area on screen to consider as a face (in pixels). |
;Maximum face size | ;Maximum face size | ||
| − | : | + | : Maximum needed area on screen to consider as a face (in pixels). |
;Analyse gender | ;Analyse gender | ||
| − | : | + | : Enable if you want to analyse gender. |
;Analyse age | ;Analyse age | ||
| − | : | + | : Enable if you want to analyse age. |
;Analyse happiness | ;Analyse happiness | ||
| − | : | + | : Enable if you want to analyse happiness. |
;Analyse angriness | ;Analyse angriness | ||
| − | : | + | : Enable if you want to analyse angriness. |
;Analyse sadness | ;Analyse sadness | ||
| − | : | + | : Enable if you want to analyse sadness. |
;Analyse surprisedness | ;Analyse surprisedness | ||
| − | : | + | : Enable if you want to analyse surprisedness. |
; Show face size | ; Show face size | ||
| − | : | + | : Enable if you want to see the face size (area occupied) on screen. |
; Show resolution | ; Show resolution | ||
| − | : | + | : Enable if you want to see the video source resolution. |
; Show frame rate | ; Show frame rate | ||
| − | : | + | : Enable if you want to see the video source frame rate. |
; Search eyes | ; Search eyes | ||
| − | : | + | : Enable if you want the system to search by eyes for a more accurate recognition. |
; Search nose | ; Search nose | ||
| − | : | + | : Enable if you want the system to search by nose for a more accurate recognition. |
; Search mouth | ; Search mouth | ||
| − | : | + | : Enable if you want the system to search by mouth for a more accurate recognition. |
; Track faces | ; Track faces | ||
| − | : | + | : Enable if you want keep tracking faces even if their data is not stored. A tracker is activated that tries to track the faces that were lost by the face detection module. |
; Time base | ; Time base | ||
| − | : | + | : Related with fps. The smaller this value is, the greater the value of fps you'll get (if your system can handle it). 25fps would require a value of 0.04 (40ms). |
; Thread count | ; Thread count | ||
| − | : | + | : Set number of threads used for processing. If you run this system on a machine with more than one single cpu (or core), it typically will be faster to use all cpus. The number of threads is limited to 10. But it only makes sense to provide at most the number of available cpus available. |
; Model type | ; Model type | ||
| − | : | + | : Defines which model is used to detect faces. For the value "Front", a model is used to detect upright frontal faces. For the value "Rotated", the detection model used is able to detect -60 to +60 degree in-plane rotated faces. For the value "Profile", the model used is able to detect upright frontal and out-of-plane rotated faces. |
; Memory length | ; Memory length | ||
| − | : | + | : An identity number is assigned to each face, an 'id', which is constant as long as the trajectory can be determined without discontinuity. This option defines for how long (in seconds) a face is remembered internally, so that a connection of a currently detected face to the same face detected some time ago can be made. |
; Memory type | ; Memory type | ||
| − | : | + | : Only takes effect if "Memory length" > 0. For "Spatial" the face id is remembered internally and assigned again if a face is lost and found again at approximately the same position in the image. However if a face is lost and found again in a different position a new id is assigned. This is the fastest method of the three. |
; Phantom trap | ; Phantom trap | ||
| − | : | + | : Tries to find false detections (phantoms). If "Off", this module is not used. If "Delete", false detections are identified and deleted on the fly. If "Mark", identified false detections are marked by the module. Each object will have an additional attribute with the key ’Phantom’. No objects are deleted in this mode but their data is stored. |
; Minimum face score | ; Minimum face score | ||
| − | : | + | : This is the minimal face detection threshold. Each face has a rating with key ’Score’ that indicates how likely it is to be a face. Using this parameter you can set a minimal score for the faces that will be returned by the engine. The lower this value is the more faces but also false detections will be found. A good value is 9. |
; Auto start engine | ; Auto start engine | ||
| − | : | + | : Enable if you want to start the engine (store data) whenever the application starts. |
; Show log | ; Show log | ||
| − | : | + | : A log window appears on the bottom. |
; Web server port | ; Web server port | ||
| − | : | + | : Port used in the Web Server. |
; Database port | ; Database port | ||
| − | : | + | : Port used in the Database. |
| − | + | ||
==Start Engine== | ==Start Engine== | ||
| − | After click the "Start Engine" button, the system will start collecting the consumer's data, storing them in a Database. | + | After click the "Start Engine" button, the system will start collecting the consumer's data, storing them in a Database. Whenever you want to view/analyse the data stored in a graphical mode click "Retail Analytics & Targeted Advertising". Please refer to [[Retail Analytics/Statistics|Statistics]]. |
Latest revision as of 16:25, 13 February 2013
Contents
Run
Getting Started
This is Retail Analytics main window, where you can start the software engine. In the left side, you have "Video Source". Select the camera you want to use with the system. Click the play button on the right. Below this, there's "Video Settings". This is where you can configure some properties of the video source, like brightness or contrast. In the right top corner, you can select "Full Screen" to run the system in full screen. To configure your triggers please refer to Triggers.
Advanced Settings
Engine Mode
There are two modes for the engine: "Kiosk" and "Big Space". The differences between the two are related to the trigger activation.
- Kiosk
- In Kiosk mode, the trigger is activated as soon as a person stands in front of the camera. The system analyses the person's data and starts its trigger, for the respective kind of consumer (by gender, age group, ...).
- Big Space
- "Engine Settings" has an option named "Big space tick frequency". This option is defined in minutes and refers to the time interval of analysis. If set to 30 (minutes) this means that, for 30 minutes the engine will analyse the consumer's data (gender, age group) without any trigger during that time. After 30 minutes and after find the predominant kind of consumer, the specific trigger will start.
Engine Settings
- Learn time
- After recognize a face, this is the amount of time spent analyzing data (gender, age, ...) that is stored in Database.
- Minimum face size
- Minimum needed area on screen to consider as a face (in pixels).
- Maximum face size
- Maximum needed area on screen to consider as a face (in pixels).
- Analyse gender
- Enable if you want to analyse gender.
- Analyse age
- Enable if you want to analyse age.
- Analyse happiness
- Enable if you want to analyse happiness.
- Analyse angriness
- Enable if you want to analyse angriness.
- Analyse sadness
- Enable if you want to analyse sadness.
- Analyse surprisedness
- Enable if you want to analyse surprisedness.
- Show face size
- Enable if you want to see the face size (area occupied) on screen.
- Show resolution
- Enable if you want to see the video source resolution.
- Show frame rate
- Enable if you want to see the video source frame rate.
- Search eyes
- Enable if you want the system to search by eyes for a more accurate recognition.
- Search nose
- Enable if you want the system to search by nose for a more accurate recognition.
- Search mouth
- Enable if you want the system to search by mouth for a more accurate recognition.
- Track faces
- Enable if you want keep tracking faces even if their data is not stored. A tracker is activated that tries to track the faces that were lost by the face detection module.
- Time base
- Related with fps. The smaller this value is, the greater the value of fps you'll get (if your system can handle it). 25fps would require a value of 0.04 (40ms).
- Thread count
- Set number of threads used for processing. If you run this system on a machine with more than one single cpu (or core), it typically will be faster to use all cpus. The number of threads is limited to 10. But it only makes sense to provide at most the number of available cpus available.
- Model type
- Defines which model is used to detect faces. For the value "Front", a model is used to detect upright frontal faces. For the value "Rotated", the detection model used is able to detect -60 to +60 degree in-plane rotated faces. For the value "Profile", the model used is able to detect upright frontal and out-of-plane rotated faces.
- Memory length
- An identity number is assigned to each face, an 'id', which is constant as long as the trajectory can be determined without discontinuity. This option defines for how long (in seconds) a face is remembered internally, so that a connection of a currently detected face to the same face detected some time ago can be made.
- Memory type
- Only takes effect if "Memory length" > 0. For "Spatial" the face id is remembered internally and assigned again if a face is lost and found again at approximately the same position in the image. However if a face is lost and found again in a different position a new id is assigned. This is the fastest method of the three.
- Phantom trap
- Tries to find false detections (phantoms). If "Off", this module is not used. If "Delete", false detections are identified and deleted on the fly. If "Mark", identified false detections are marked by the module. Each object will have an additional attribute with the key ’Phantom’. No objects are deleted in this mode but their data is stored.
- Minimum face score
- This is the minimal face detection threshold. Each face has a rating with key ’Score’ that indicates how likely it is to be a face. Using this parameter you can set a minimal score for the faces that will be returned by the engine. The lower this value is the more faces but also false detections will be found. A good value is 9.
- Auto start engine
- Enable if you want to start the engine (store data) whenever the application starts.
- Show log
- A log window appears on the bottom.
- Web server port
- Port used in the Web Server.
- Database port
- Port used in the Database.
Start Engine
After click the "Start Engine" button, the system will start collecting the consumer's data, storing them in a Database. Whenever you want to view/analyse the data stored in a graphical mode click "Retail Analytics & Targeted Advertising". Please refer to Statistics.
