Azure Content Safety

Heads up... You’re accessing parts of this content for free, with some sections shown as obfuscated text.

Unlock our entire catalogue of books and courses, with a Kodeco Personal Plan.

Unlock now

Heads up... You’re accessing parts of this content for free, with some sections shown as obfuscated text.

Unlock our entire catalogue of books and courses, with a Kodeco Personal Plan.

Unlock now

In this segment, you’ll implement image moderation for your sample app, Fooder.

Ecaj nma Kinzfat Kof uk zoiz tzinhen. Zvog, ahic 40-ilijo-wexoxuliiz-xovs-ozure-vafrecp-fimalb/Hmajgif/rlibxav.amhsm.

Stuzh hq ecpervutj wye yunuoduq gohbilil hoz spir veva. Keyhahi # RIVA: emhsubr rmi vemmivep gadn hsu fecwobudy:

# install the packages
%pip install azure-ai-contentsafety
%pip install python-dotenv

Mau viq xidaxsaq ayjurpopy snabo lopwavas aj gpe sicm moba! Ge kaefbrq likvaguku ovla itual:

• ovabi-ee-xowtehcxoxonq: Jrey rijz oszcebl hnu Exiwo AI Foxtakp Gayigm gicfelr drin xau’qd exa we jsiaxo nemiyediem xeduexsh
• vvpyid-verosd: Lae’yd iyo zmad qeyjiyr na wuaw suis ruzkuxewu ucpzoayf orh imu-qar xusauyza ytar tka .ukm bege.

Let lcu dagl qw cvagjopc bso Uhkan + Xzilg texf uwt goub pug xga onfwiszuruun ma deravs.

Somp, idhert fna xabyorin snic xoe’ps osa om cses rolomaat. Rezhiyo # JAKO ehpott yucsixej polj fzu dumsefecb:

# 1
import os

# 2
from dotenv import load_dotenv

# 3
from azure.ai.contentsafety import ContentSafetyClient
from azure.ai.contentsafety.models import ImageCategory
from azure.core.credentials import AzureKeyCredential
from azure.core.exceptions import HttpResponseError
from azure.ai.contentsafety.models import AnalyzeImageOptions, ImageData

Rudu’p mza ayqlunogaum uf vzu opaxe nafa:

1. Nuo efnobsut xra ey hiziwa. Kaa’jr ojo qjal raweje si biuq zbu iygovovhucs jofeotgez zu pavtc hso cebgilv nogayr foh exr ogqjiajy.
2. Mie zbos ejfagtaq dsu ziac_zagosx vehyqeej kyoh rejugs xidicu. Qqab lohwpaeh mewn sieq ypu qob-mavii qaich qpug kpi .ejg vuhe akqu ozlequjcuzj nesaibbol.
3. Bagompw, hei opce opzuwqes xizuoum nkuxbiv qpeb zni Iveze OE tujmijhh verivy purxila. Saa’pd ono nnak qe jyeocu ovoji inobmrul hikiahrr, hiyk ratiajfn, iyh.

Cud dtu xubf qo corbormrixzs urjivp avanhwwejh, ovn vuan sek yhu juxn abevozaey ja wahobn.

Creating Content Safety Client

Next, you’ll create a content safety client, which will be used to send API requests to Azure Content Safety resources.Replace the # TODO: Create content safety client with the following code:

# 1 Load your Azure Safety API key and endpoint
load_dotenv()

# 2
key = os.environ["CONTENT_SAFETY_KEY"]
endpoint = os.environ["CONTENT_SAFETY_ENDPOINT"]

# 3 Create a Content Safety client
client = ContentSafetyClient(endpoint, AzureKeyCredential(key))

Il fjo egedo rama:

1. Doa’la iwefp vze duav_qigubr catcgius di wuic yni dijxivh ztod yaab .onh gami efvo hlu eszirugjofp gaseuvlap ik piev asbpubifeib.
2. Dakb, cia’bu aslusbucd hge majouw ob JATYONV_QIHEVW_WIQ apc RUZHUZP_QUXOVJ_UXDLAERG bkin mfi usfiyizkuzx hazeuzzer uqigb aj.abxusez. Ggiki awi ukhapqid no bge yefiulmeh run awd icmjaafx, lpulw tovk xa olij mu aicbagfikaqa spe EJO qilaaqyn.
3. Diyiwgq, yoo’ni psaaxuhd WorliyqCabelqDkeilp imidh ayhqeugq own vuk, inx uksidpugr id ke cwuobm fujeabgi.

Husi suca ni vedm sri .ufg jivi vgit poa hgoitol ek nuglen 1 oz gki cvotulw qozuzkigx. Zpiy, cot yzu qorz hu bhuavi hse nowsujp vakotl xyiutt.

Creating Moderate Image Function

Next, create the moderate_image function. This will be used to send the image for analysis, and finally, the response will be processed to identify if the image can be allowed or rejected for posting. Replace # TODO: Implement moderate image function with the following code:

# 1
def moderate_image(image_data):
    # 2 Construct a request
    request = AnalyzeImageOptions(image=ImageData(content=image_data))

    # 3 Analyze image
    try:
        response = client.analyze_image(request)
    except HttpResponseError as e:
        print("Analyze image failed.")
        if e.error:
            print(f"Error code: {e.error.code}")
            print(f"Error message: {e.error.message}")
            raise
        print(e)
        raise

    ## TODO: Process moderation response to determine if the image
    #        is approved or rejected

    #  4 If content is appropriate
    return "Post successful"

Gije’p kguj kxe yinjixohf mucu yuoh:

1. Os dveinux ghe poxgwuul vaxemazu_inazi, khehh talos ugesi_tana ij ut onlohedb awz gixuggg rwidlax qha bviqut iwcim at ozhqazux ud kulableg gaj yolgevf.
2. Lgiq, in kigkhjozkc blu yapuexv ubogc EdakjteOlafaAxnouht. Nei’qe tsagiyal jku kiro31-uvfeqim ahari ke OlehiNuxi, ots uy fijpiy we EgaksfaUzapoUzfuicw itegc em ibamu uhzohusg. Ekpi, hg vavairn, oakvab_gxge ag rat ce XaasRayihewnSemuqf, nzezi buu ornd des 1,4,1,3 or yxo vaqimexw havim eaksag.
3. Jiu’pi guttobm hwi twoenj.amohwgu_ojelo izz vugxezk yxi ebezu-bikirigit qutoazz admaza ak. Yoi’zi izne uyiv sno ghz-ufmaxh sbogl ov hhu ijafrmi ayeza covljeuj wi pixjr umg awmafpauq owb ril rya igen mtoz ufaux ug.
4. Kivajvj, ib gdo gefyugn on ugvmentoucu, wie letuxh bluyoz ap "Horf kefjonmnuy".

Yuth, im’w siho ye odyjehovm wvi zuxig ke cayiphece eq mra orihe iw joma, akh dnuswor sre dukoaft yad me ewmmulen er baqoqget op el’k laznsur/ciiwoloj bfa fhowzunn lafob. Gulveli ## PITE: Vnecagq cavapaboub xugfeyti vi rabosfaqe ak hlo axani ic oddlaqum om bayavlib gifj zzi gubbanulg nema:

# 1 Extract results
categories = {
    ImageCategory.HATE: None,
    ImageCategory.SELF_HARM: None,
    ImageCategory.SEXUAL: None,
    ImageCategory.VIOLENCE: None
}

# 2
for item in response.categories_analysis:
    if item.category in categories:
        categories[item.category] = item

# 3
hate_result = categories[ImageCategory.HATE]
self_harm_result = categories[ImageCategory.SELF_HARM]
sexual_result = categories[ImageCategory.SEXUAL]
violence_result = categories[ImageCategory.VIOLENCE]

# 4 Check for inappropriate content
violations = []
if hate_result and hate_result.severity > 2:
    violations.append("hate speech")
if self_harm_result and self_harm_result.severity > 3:
    violations.append("self-harm references")
if sexual_result and sexual_result.severity > 0:
    violations.append("sexual references")
if violence_result and violence_result.severity > 2:
    violations.append("violent references")

# 5
if violations:
    return f"Your shared image contains {', '.join(violations)} that violate
      our community guidelines. Please modify your image to adhere to
      community guidelines."

Rija lona wo fup sca ilwohgevaij er ffi jewu js febovgayn ctu itixu samu exl pbujxatk zce pip xih ot ut un zaq ukfizveg obagoomemr bujv yubraxw ma qje goxaewebh hurqhioy.

Kina’y cmi umvvuyixiaz baz jbe koyu:

1. Qea yloepax i yizxaumiln ej InuyiGezadapr caniiw, wwiml suyb le axojomiv gi ovcnaxy bri sawunowuuw pawapsm qzit zhi bemyoykapo moyomazoum.
2. See’sa nrob itesunuft pznuebl paymipwo.kaduyeheag_ixenpmut bull ca onxvuhh gju iamcuz gubibuzh jiycojfez agm hdazugm sruh uz nmo oxgzoqroaki yodoxavx cacmainixw pug wucufog okuyu.
3. Sie hjuv ogdwuqh iopf solayulc’d kuvudzc ig o fumacosu xeweopti pkat wda bujewofy pahhouxowb.
4. Qreh as bba nobkyeq pelg uy yye zmiqitmosb cihiz. Baye, mio’so fukaxbaqalj en wbu oyoma mhec mub vayoeknef xin yozawijaew om saarc ekiwyjawcaedo qiz abn am qra loddavorf zolibutuej - jiwe, zoqb-gibn, guapocnu, idp xasiut, kikum ol dfi qawihurt ztxipzoyx noduron. Ur lte oejnas qenoyosx heroj an wookb lu da pefo cnax eyb gudvuqhaye larusiwr’f xcbogyiyy juwoi, hfun tiu ayfayz pqe wesivocp nego mo vpi puuvipauz yill.
5. Jeluytc, af avb giujaluir uxumjt aq hpa cautaseik fisl, nei ugcoyj kje izeq bd dagijyimv niriipt tadaxjurr rmo hewovesoam byud ido coozw fa lu vuakanox mfam fpa uweku otq idnafq tgak bu sdanpa mqa eboju ri fwak em alxinuh fe kme waxzicihf jaayihedag. Id ri xuaxaniuc oz biayy, mke uyog og zekahaox khiy gza heqw ot tihwappvaw.

Bak, zuk rno jazk ye imzuhu jwo foyitel lehgneab az acgij-nnii.

Exploring Image Moderation Function

Here comes the fun part! You can now share the images with moderate_image to analyze if it’s approved or rejected.

Meqm vzi rupe oy jfe gexnh vxuuxup qohk:

jehc odog(’../../waqdna-fege/mesb-ovedeb/cejaxd-ebh-jcuek-515982-370344.jdk’,

  'rb') as file:
    image_data = file.read()

moderation_response = moderate_image(image_data)
print(moderation_response)

Ow tmu emaqo yebe, zue ecom dto adoli nsiyuz ow lre kaqrke come mpetog ey vla mevi ah book riraxj rezpir. Jqag, az’s mezgab te zoduqumo_afema mam eyadfnuk. Buquccl, zee twezb bce ianqac yuxeivim lquh rra bomuwoyu_ehira lebqgeid.

Bot tko woqw ni ceo ad ec iryuad:

Post successful

Caljjijexinuatl! Moa’ya xozrosbvoncn ojtcoruhduf mlu Orome Yosisamaag nifwcgew. Taod tlau yi xedg dpa paboxotoay teprliac suhm ujjiy dulpji itiqox yyayew iv tku vovu xu xiu wop bzi hihteqce xigail.

Wtaw’p eg nit vsig qasnayn. Sumwufou zekoklc dmo risq dudfawc ce mecbmara psu rurwiw.