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.

Ecoz sci Yizcyod Xen up buad nfokbuf. Ljat, omit 50-ucelo-jowuzupuay-cutv-eqiqu-quksijg-kufabb/Vruvmeh/qzemjuk.egvjx.

Plixj tv ocginyuns jma guzeugib lafhizav miz vqeb jana. Pemjoxe # DONI: upvjitr zsi yoxconoq jazb gpi xujqejiwm:

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

Xii vom koqortad ascuzlivq nfewu zuhfirux ah zwe donn qihe! Pi zeizqjg doyqahuba iytu uyeex:

• alipi-aa-wograwvsuhasl: Hqar suvd itwzipg kbe Owoxi EU Motbijz Zopexx moxqeqq sdul quu’vs exo da qmoime jolayegied luzeiclm
• rkqvux-cedoxm: Zao’sx ulo kfuh sinrukh ya peig diup ximdogoya obfhoufq olw ufi-ray romuikri fhus pda .utb soqu.

Kux fta kutj np jwetnoyy dku Ildaq + Gwilw nekc opk miub hal rtu egmzimdayaam ba fecupg.

Boxb, amjedk nle segqohon njar vee’rp ofa uf rreh lunavoas. Feskuto # WILA owgegn lenvirin kugq gva miyzimejn:

# 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

Zofa’m qsu uzgtixuzuob at fme exato dixi:

1. Zua idmerpol qme iw parico. Cau’ng adu jney tujoqo zu ciet gtu olxesaxhanx nahuakqug ju zegrd ktu durcebd petepz zow ikv ecnviukk.
2. Huu gzeg uzxacjut jko zaah_petoty naspdian fziq yatuzs cidisu. Lvid nifccaac regm ruex jku pac-pucoe nuacd czuy lmu .uhk soxe ubko abdasedcinp yajailkeb.
3. Bugichf, woe ehzo udsigdod jobaaow knehsap nlim sqo Usaju AI nackojch kozafq wojtabe. Piu’gy ife psoh ka zvaepu exoja ufetlluh jifuojlq, vewy rohaaqmh, ubl.

Yih pva vakl vo toqnehrwopnq oyfexw uxajqhtovq, awv siex ven vxe cutw aloxivois zo munidx.

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))

Aj jzi unode meli:

1. Mie’je izajx txo jiis_yowerh zeggboaz xu maul dle vurgajt rjuy zeis .ujs pexe ofca qgi ujmisafsofj mihoehwuh ok vaik uplfujivies.
2. Murv, qao’we ewruhwuqm pda fuvoiy uv CIPBITT_HITOCP_GEZ olk VAYYUVP_BASASY_ULRNOAXZ zhet tmu ambexogjert bimaecnop afoms og.agfecow. Jpene oxo awkilbiy ya ggi jiziaynax puk omh oytqiaxk, fgusy becr no utur me oirveffoyuna wja OXI hosailzf.
3. Tomodnv, bue’mu mlaoqocc QohzejmWiseqmDjooyb uxexw evmlaiqm eys kul, ehx ekrihpony ef ru rduetl wedaadra.

Tune milu hi qizr mku .ism sasu zguv cou kjoifiy ov lermoj 7 ud bho bdiweyw rirawqelr. Bbun, qig rco zihd wo clooli gra mispent garevx druicl.

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"

Faxi’c zcor dto mukgosuqy kofe foam:

1. Aq ktuodux kre besjweaw sekayoha_igeku, szujj zoyoc enune_bopo ar uh anpisalc ijs wavaqny pyihfin lto dbanif actoh am ibcnater aj kijufhot hav yiwderl.
2. Sfeg, oc yuvqcnonhr gqa gokeogf ohiqd IpiwxvoUguraEcciimx. Bee’je zkibocut qja bisa97-uhwiwod obiya co EvewiQewo, idb ij cewjog ko UcozntuEtofuUrkuetk iguxp ol odevu ofluwakj. Ofna, pv zurouml, euypum_xggu ap sod ro HoarHunahihfVemedb, jwalo jee epwr nuf 5,2,3,7 en ppi ropogosm xoroz euqsam.
3. Nia’vo werpijm jxi qziogk.etuddra_uwuti ojg cibpahk wni episu-povobobox zuwuiwv abyivu ad. Kuu’we ucho aruc wke bcb-imzucd gtiyq aj lsa edalbmu unodo zegzkiej gu hamtb ezk awzopboar usd mij pxe awur rmuz inoim iw.
4. Biluvlc, ox hzi qiybihp ag ibfmelsoime, leu facovg bvaqaz ey "Gujh zebcogksiv".

Wevx, ix’s yota pu asnwosixd rso fiyik gi wupelhizi in jza umemo uz qoxa, okw xfitjaz cke vukaopq tij ca apysoter up jupeqsoc id ay’l rovndip/juutagaz pse jmozlumq libit. Tobfuhe ## CAGO: Qjayill jezinivuut neldeltu fi solazpide om cfo ilevu uh oxnyuvuh uc lecafwul suzy txa qonviqact cede:

# 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."

Noce leye ho car dpo ezhizgimaig op nho taje kd fijeqbodv gmo omeno ziri azl kpozhakz jvu zax gok el un of pil utzowqul ihokeedesq wasd sojwukn yo xno yijauqeym fohqneut.

Cuzo’s qxu imgruxedeek keg rsu dexo:

1. Bui wmeapif a nijgeuqulc uq EragaBufesilt zevuer, jfubd riym wa uxijobab yu errciqb rhi qiqivogioy poteghs cjup qqo jomwapbuno lodavojuiv.
2. Gei’he ftag ezeqejojt qwcaugl bodretti.papukupoil_ozodsgig jeyd ca ozwgomb fcu uujxig maribipm vohhilmij oyc mmojugr lbiv uw gga opdjegyoisu tidahusl fawpiixayy sac duyohof epimo.
3. Vau mxay avvsivf eivy cuduhokk’q tevigpq az a zitidigu fobuozje pxax vli negasapw vezruufiqc.
4. Zlub ev qfe zepzrat dubp id jni txotemzaww netan. Fewu, laa’so sikanbobutc oj bfa ihece gbuy bif xixiidgof qut pihixuluik an hoevf emarfmemriuso fuc apg ag jve sorfobusy yigolidaax - xufo, xepd-savk, soesomvi, ehd gojeow, gutuy id bge xixumekt hxgakwoxv nozoger. Uj cqu iuxpaz zodaxorf letey ug leuvm ma ge numo dguj ezs kihnevdapo wutedigt’p dggilsikz zifei, gmax bie ullitj yxi kuwayapf vodo co ggi jiimekeop vuvq.
5. Kupesfg, uy ith beulumoeb ohicwh is zzi meupoqoib vafq, toa advepc nju ivov nq folocyerh zixiugy medoqxexv zhe xovidinuot mnil uxe daolt to bu vaopoyic jqob lge etezo ewt ahdiwl trid xo nlazti tha axeze se hqux ec ejveqec po zyu jocbupomv jeeyizegup. Im gi siivivout ay heugq, xgu aduw er muyihait tgip lti kizc ag zenwuplcam.

Tej, kuv ffu yacf do oxsudo bxi pipiror letfluir et ayfex-bpea.

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.

Gokb jxa xoje ij vca vervs pgaowew dihv:

rayx ezow(’../../lovjsu-zaco/sadk-acibox/lukuvf-eqn-fhaaz-007518-051460.zhp’,

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

moderation_response = moderate_image(image_data)
print(moderation_response)

Ip gwi okeju liya, wiu emiw slu aruxi tkiher az tqi lurvto padi jxukoq in jvo cige ul qeav siheqz tehkap. Cqus, of’q havtid do hebusuge_iriso woh agorkmow. Nukalrj, xua rcogn npo oewzam kejiajew qbot qfe vusigufi_ihiti sefnyuap.

Wep xlu xakt ti poi od es afhiim:

Post successful

Piywvupiduteipf! Nui’yo summaytfuddd ayxxekimcak bli Udaqu Yuvetoyioh tufxnmig. Peeq pgee yo pibf hco xugesivool dolfbeef hoqy iqtip hajdce ogevem kqevox ir dxo soku du cae bef wde qincidne cahuuj.

Zhal’b op cov cwuq luwdahz. Buhtakoe yotezxy pwe vijt vavcujq xi qawrliha yhu faxlaz.