Detect Suspicious Activity Using Image Analysis

Overview

This feature helps users to identify the suspicious behavior of candidates by monitoring a candidate’s webcam activity during the test duration. Through this, users can identify, review, and eliminate possible cases of cheating.

The captured images will be categorized under different sections based on the category under which an image is flagged by the HackerRank Image Analysis solution. Five images will be displayed under each section for a quick preview of the suspicious activity.

For example, up to five candidate images having multiple faces can be found in the Multiple Faces sections under the Proctor tab of a Test attempt.

This way, the Hiring Managers or Recruiters can quickly verify the candidate's attempt for suspicious activity and move on with their next steps in the recruitment process. With this change, we will be optimizing the hiring process and creating efficiencies by saving time for internal teams.

Enabling Image Analysis

Image Analysis is turned off by default on the Test Integrity page. You can enable the feature after consulting with your team.

1. Under Image Proctoring, enable the Image Analysis toggle.
2. You can click on the help icon to view the related support article. You can also view the HackerRank NYC laws for AI in the article FAQ section of HackerRank Services and AI Laws, by clicking the link here.
   Additionally, HackerRank's AI-powered plagiarism solution has gone through an independent bias audit consistent with the NYC AI Law. Summary of Bias Audit Results (HackerRank's Image Analysis System)
3. When you enable the Image Analysis detection, the following prompt displays the effects on your workflow:
4. Provide your consent and select the check box.
5. Click Agree & Enable; Image Analysis will be used to detect any suspicious activity of a Candidate.

Candidate Experience

When you enable Image Analysis, the system automatically takes a Candidate's consent to proceed further. The Candidate will be notified with a prompt message to allow webcam access.

1. When the Candidate launches the Test, a prompt message is displayed to provide a consent that the Candidate is fine with taking the proctored  test and allow the web cam access.
2. After allowing the web cam access, the Candidate is notified to take the Test in a well-lit and a private place.
3. When the Candidate clicks Continue the web cam captures the Candidate's image to confirm that the same Candidate is taking the Test.
4. Once the Candidate is allowed access to the Test, a confirmation is prompted to capture the candidate's biometric information.
5. The Candidate can agree to the declaration and click Agree & Start to start taking the Test.

Note: Candidates can take Proctored Tests only using Chrome or Firefox browsers.

Reviewing a Candidate’s Suspicious Activity

You can review the suspicious activity of the candidate by going to the detailed report from the Candidate list page of a Test.

1. Goto a Test and click the Candidates tab. The Candidate list is displayed.
2. Click on a candidate to view the report. The Summary Report page is displayed.
3. Click the Proctor tab to view the Image Analysis of the Candidate captured during the Test attempt.

Reviewing Candidate’s Suspicious Activity on the Detailed Report Captured on Webcam

When a Candidate attempts a Test with the Proctoring of Image Analysis enabled for that Test, then a series of images of the Candidate are captured during the Test. These images display any suspicious activity detected during the test. The images are categorized based on the following:

Multiple Faces

The illustration displays the experience captured when multiple faces are detected during the Test. The user can then assess, based on these images, how to proceed with the candidate.

Secondary Faces

The illustration displays the experience captured when secondary faces are detected during the Test. This helps to identify any imposters trying to take a test on a candidate's behalf. The user can then assess, based on these images, to proceed with the candidate.

No Faces

The illustration displays the experience captured when no faces are detected during the test. This helps to identify if a candidate is trying to take help or if there is any other malpractice. The user can then assess, based on these images to proceed with the candidate.

HackerRank also collects feedback from users to ensure that we improve the precision of our solution.

Reviewing Suspicious Webcam Activity on the CSV Exports

The proctoring information is also captured on the exported CSV files:

• In the CSV exported file, three additional columns are added to make it easy for the users to filter out Candidates based on the suspicion severity, suspicion score, and the reason for suspicious activity, as shown in the following table:

These examples help you to filter candidates based on their Proctoring activities.

Example 1: If you are concerned only about highly suspicious candidates, use the Proctor Activity Severity column and filter out candidates marked as High.   

Example 2: If you are fine with moving ahead with Candidates flagged for reasonable suspicious activity because you do not have enough candidates, use the Proctor activity suspicion score column to filter candidates appropriately.

For example, you may consider progressing candidates on the lower end of the suspicion scale. As Medium ranges from 5-7, you can use the Proctor activity suspicion score column and choose candidates marked as 5 (5 being the lowest on the Medium scale).

Note: Suspicion scores between 8-10 are marked as High, 5-7 are marked as Medium, and <5 are marked as No.

Recommendation: When you progress candidates marked as a Medium to the next stage, it is recommended that you assess the candidates further to ensure their skills are a good match and that the candidate did not cheat in the previous round.