Porn Detection Event Notification

Recent Pages

Porn Detection Event Notification

Last updated: 2021-11-09 09:37:47

CSS porn detection takes real-time screenshots of suspiciously pornographic video images in live streams and stores them in COS. The porn detection callback is used to push the information of detected pornographic images, including their type, score, and screenshot time. You need to configure the server address for receiving porn detection callback messages in the callback template and bind the template to the push domain name. When a porn detection event is triggered by a live stream, the Tencent Cloud CSS backend will call back the pornographic image information to the configured receiving server.

This document describes the fields in callback message notifications sent by Tencent Cloud CSS after a porn detection callback event is triggered.

Note

You need to understand how to configure callbacks and how you will receive messages via Tencent Cloud CSS before reading this document. For more information, see How to Receive Event Notification.
By default, only questionable results of porn detection will be called back.
We recommend you use the [type](#type) of an image to determine whether it is pornographic. As the detection results are not 100% accurate and there may be false positives or false negatives, you can confirm them manually if necessary.

Screencapturing Event Parameters

Event type parameters

Event Type	Parameter Value
CSS porn detection	event_type = 317

Common callback parameters

Parameter	Type	Description
t	int64	Expiration time, which is the Unix timestamp when the event notification signature expires. The default expiration time of a message notification from Tencent Cloud is 10 minutes. If the time specified by the `t` value in a message notification has elapsed, it can be determined that this notification is invalid, thereby preventing network replay attacks. The format of `t` is a decimal Unix timestamp, i.e., the number of seconds that have elapsed since 00:00:00 (UTC/GMT time), January 1, 1970.
sign	string	Event notification security signature sign = MD5(key + t). Note: Tencent Cloud concatenates the encryption key and `t`, calculates the `sign` value through MD5, and places it in the notification message. When your backend server receives the notification message, it can confirm whether the `sign` is correct based on the same algorithm and then determine whether the message is indeed from the Tencent Cloud backend.

Note：

key is the callback key in Event Center > Live Stream Callback, which is mainly used for authentication. We recommend filling in this field to ensure data security.

Callback message parameters

Parameter	Required	Data Type	Description
streamId	No	String	Stream name
channelId	No	String	Channel ID
img	Yes	String	Link to the alerted image
type	Yes	Array	Returns the maliciousness label with the highest priority in the detection result (labelResults), which represents the moderation result suggested by the model. We recommend you handle different types of violations and suggestions according to your business needs. Returned values: 0: normal 1: pornographic 6: abusive 8: advertising 2–5 and 7: other offensive, unsafe, or inappropriate types of content
score	Yes	Array	The score of `type`
hotScore	Yes	Number	The score for a sexy image
pornScore	Yes	Number	The score for a porn image
illegalScore	Yes	Number	The score for an image with illegal content
polityScore	Yes	Number	The score for an image with politically sensitive content
terrorScore	Yes	Number	The score for an image with terrorism content
abuseScore	Yes	Number	The score for an abusive image
teenagerScore	Yes	Number	The score for an image with content inappropriate for teenagers
adScore	Yes	Number	The score for an advertising image
ocrMsg	No	String	OCR text (if applicable)
suggestion	Yes	string	Suggestion. Valid values: Block Review Pass
label	Yes	string	Returns the maliciousness label with the highest priority in the detection result (labelResults), which represents the moderation result suggested by the model. We recommend you handle different types of violations and suggestions according to your business needs. Returned value: Normal: normal Porn: pornographic Abuse: abusive Ad: advertising Custom: other offensive, unsafe, or inappropriate types of content
subLabel	Yes	string	Sub-label name. If no sub-labels are hit, an empty string will be returned
labelResults	No	Array of LabelResult	Moderation result of the category detection model, including pornographic, sexy, terrorism, illegal, and other results
objectResults	No	Array of ObjectResult	Moderation result of the object detection model, including politically sensitive object, advertising logo, QR code, and other information
ocrResults	No	Array of OcrResult	Moderation result of the OCR text, including OCR text and text moderation result details
libResults	No	Array of LibResult	Moderation result of the risky image library
screenshotTime	Yes	Number	Screenshot time
sendTime	Yes	Number	The time the request was sent in UNIX timestamp format
similarScore	No	Number	Image similarity score
stream_param	No	String	Push parameter
app	No	String	Push domain name
appid	No	Number	APPID
appname	No	String	Push path

LabelResult

Hit result of the category model.

Name	Type	Description
Scene	String	Returns the scenario identified by the model, such as advertising, pornographic, and harmful.
Suggestion	String	Returns the operation suggestion for the current maliciousness label. When you get the determination result, the returned value indicates the action suggested by the system. We recommend you handle different types of violations and suggestions according to your business needs. Returned values: Block Review Pass
Label	String	Returns the maliciousness label in the detection result. Returned values: Normal: normal Porn: pornographic Abuse: abusive Ad: advertising Custom: other offensive, unsafe, or inappropriate types of content
SubLabel	String	Sub-label name
Score	Integer	Hit score of the label model
Details	Array of LabelDetailItem	Sub-label hit result details of the category model

LabelDetailItem

Sub-label hit result details of the category model.

Name	Type	Description
Id	Integer	ID
Name	String	Sub-label name
Score	Integer	Sub-label score. Value range: 0–100

ObjectResult

Object detection result details.

Name	Type	Description
Scene	String	Returns the identified object scenario, such as QR code, logo, and image OCR.
Suggestion	String	Returns the operation suggestion for the current maliciousness label. When you get the determination result, the returned value indicates the action suggested by the system. We recommend you handle different types of violations and suggestions according to your business needs. Returned values: Block Review Pass
Label	String	Returns the maliciousness label in the detection result, which represents the moderation result suggested by the model. We recommend you handle different types of violations and suggestions according to your business needs. Returned value: Normal: normal Porn: pornographic Abuse: abusive Ad: advertising Custom: other offensive, unsafe, or inappropriate types of content
SubLabel	String	Sub-label name
Score	Integer	Sub-label hit score of the scenario model. Value range: 0–100
Names	Array of String	List of object names
Details	Array of ObjectDetail	Object detection result details

ObjectDetail

Object detection result details. When the detection scenario is a politically sensitive object/figure, advertising logo, QR code, or face attribute, it represents the label name, label value, label score, and location information of the model detection frame.

Name	Type	Description
Id	Integer	ID
Name	String	Label name
Value	String	Label value: When the scenario is `Ad`, it represents the URL. For example, when `Name` is `QrCode`, `Value` will be `http//abc.com/aaa` When the scenario is `FaceAttribute`, it represents the face attribute information; for example, when `Name` is `Age`, `Value` will be `18`
Score	Integer	Score. Value range: 0–100
Location	Location	Detection frame coordinates

Location

Coordinates.

Name	Type	Description
X	Float	Horizontal coordinate of the top-left corner
Y	Float	Vertical coordinate of the top-left corner
Width	Float	Width
Height	Float	Height
Rotate	Float	Rotation angle of the detection frame

OcrResult

Detailed OCR result.

Name	Type	Description
Scene	String	Indicates the recognition scenario. Default value: OCR (image OCR).
Suggestion	String	Returns the operation suggestion for the maliciousness label with the highest priority. When you get the determination result, the returned value indicates the action suggested by the system. We recommend you handle different types of violations and suggestions according to your business needs. Returned values: Block Review Pass
Label	String	Returns the maliciousness label with the highest priority in the OCR detection result, which represents the moderation result suggested by the model. We recommend you handle different types of violations and suggestions according to your business needs. Returned value: Normal: normal Porn: pornographic Abuse: abusive Ad: advertising Custom: other offensive, unsafe, or inappropriate types of content
SubLabel	String	Sub-label name
Score	Integer	Sub-label hit score of the scenario model. Value range: 0–100
Text	String	Text content
Details	Array of OcrTextDetail	OCR result details

OcrTextDetail

OCR text result details.

Name	Type	Description
Text	String	Returns the text content recognized by OCR (up to 5,000 bytes)
Label	String	Returns the maliciousness label in the detection result. Returned values: Normal: normal Porn: pornographic Abuse: abusive Ad: advertising Custom: other offensive, unsafe, or inappropriate types of content
Keywords	Array of String	Keywords hit under the label
Score	Integer	Hit score of the label model. Value range: 0–100
Location	Location	OCR text coordinates

LibResult

Blocklist/Allowlist result details.

Name	Type	Description
Scene	String	Scenario recognition result of the model. Default value: Similar
Suggestion	String	Returns the operation suggestion. When you get the determination result, the returned value indicates the action suggested by the system. We recommend you handle different types of violations and suggestions according to your business needs. Returned values: Block Review Pass
Label	String	Returns the maliciousness label in the detection result. Returned values: Normal: normal Porn: pornographic Abuse: abusive Ad: advertising Custom: other offensive, unsafe, or inappropriate types of content
SubLabel	String	Sub-label name
Score	Integer	Recognition score of the image search model. Value range: 0–100
Details	Array of LibDetail	Blocklist/Allowlist result details

LibDetail

Custom list/blocklist/allowlist details.

Name	Type	Description
Id	Integer	ID
ImageId	String	Image ID
Label	String	Returns the maliciousness label in the detection result. Returned values: Normal: normal Porn: pornographic Abuse: abusive Ad: advertising Custom: other offensive, unsafe, or inappropriate types of content
Tag	String	Custom label
Score	Integer	Model recognition score. Value range: 0–100

Sample callback message

{
        "ocrMsg": "",
        "type": [1],
        "socre": 99,
        "hotScore": 0,
        "pornScore": 99,
        "screenshotTime": 1610640000,
        "level": 0,
        "img": "http://1.1.1.1/download/porn/test.jpg",
        "abductionRisk": [],
        "faceDetails": [],
        "sendTime": 1615859827,
        "illegalScore": 0,
        "polityScore": 0,
        "similarScore": 0,
        "terrorScore": 0,
        "abuseScore": 0,
        "teenagerScore": 0,
        "adScore": 0,
        "suggestion": "Block",
        "label": "Porn",
        "subLabel": "PornHigh",
        "labelResults": [{
                "HitFlag": 0,
                "Scene": "Illegal",
                "Suggestion": "Pass",
                "Label": "Normal",
                "SubLabel": "",
                "Score": 0,
                "Details": []
        }, {
                "HitFlag": 1,
                "Scene": "Porn",
                "Suggestion": "Block",
                "Label": "Porn",
                "SubLabel": "PornHigh",
                "Score": 99,
                "Details": [{
                        "Id": 0,
                        "Name": "PornHigh",
                        "Score": 99
                }, {
                        "Id": 1,
                        "Name": "WomenChest",
                        "Score": 99
                }]
        }, {
                "HitFlag": 0,
                "Scene": "Sexy",
                "Suggestion": "Pass",
                "Label": "Normal",
                "SubLabel": "",
                "Score": 0,
                "Details": []
        }, {
                "HitFlag": 0,
                "Scene": "Terror",
                "Suggestion": "Pass",
                "Label": "Normal",
                "SubLabel": "",
                "Score": 0,
                "Details": []
        }],
        "objectResults": [{
                "HitFlag": 0,
                "Scene": "QrCode",
                "Suggestion": "Pass",
                "Label": "Normal",
                "SubLabel": "",
                "Score": 0,
                "Names": [],
                "Details": []
        }, {
                "HitFlag": 0,
                "Scene": "MapRecognition",
                "Suggestion": "Pass",
                "Label": "Normal",
                "SubLabel": "",
                "Score": 0,
                "Names": [],
                "Details": []
        }, {
                "HitFlag": 0,
                "Scene": "PolityFace",
                "Suggestion": "Pass",
                "Label": "Normal",
                "SubLabel": "",
                "Score": 0,
                "Names": [],
                "Details": []
        }],
        "ocrResults": [{
                "HitFlag": 0,
                "Scene": "OCR",
                "Suggestion": "Pass",
                "Label": "Normal",
                "SubLabel": "",
                "Score": 0,
                "Text": "",
                "Details": []
        }],
        "streamId": "teststream",
        "channelId": "teststream",
        "stream_param": "txSecret=40f38f69f574fd51126c421a3d96c374&amp;txTime=5DEBEC80",
        "app": "5000.myqcloud.com",
        "appname": "live",
        "appid": 10000,
        "event_type": 317,
        "sign": "ac920c3e66**********78cf1b5de2c63",
        "t": 1615860427
}

Recent Pages

Porn Detection Event Notification

Note

Screencapturing Event Parameters

Event type parameters

Common callback parameters

Callback message parameters

LabelResult

LabelDetailItem

ObjectResult

ObjectDetail

Location

OcrResult

OcrTextDetail

LibResult

LibDetail

Sample callback message

Was this page helpful?

Was this page helpful?

About Tencent Cloud

Help & Support

User Center

Resources

Sign Up

Log in

Recent Pages

Porn Detection Event Notification

Note

Screencapturing Event Parameters

Event type parameters

Common callback parameters

Callback message parameters

LabelResult

LabelDetailItem

ObjectResult

ObjectDetail

Location

OcrResult

OcrTextDetail

LibResult

LibDetail

Sample callback message

Was this page helpful?

Was this page helpful?

About Tencent Cloud

Help & Support

User Center

Resources