Abstract

Artificial Intelligence continues to support measurable improvements across physical assessment processes. One such application is AI height measurement using computer vision. In Physical Efficiency and endurance tests, combining visual data with automated analysis improves the credibility of height evaluation. This approach replaces manual-only methods with a verifiable digital process that supports consistent measurement outcomes

Problem

Accurate height assessment during Physical Efficiency Tests often faces challenges. Manual measurements can introduce variation, disputes, or a lack of recorded proof. In large-scale examinations, maintaining uniform measurement standards becomes difficult, making height verification a sensitive and error-prone task.

Solution

An AI and machine learning solution was implemented to measure candidate height using images captured during Physical Measurement Tests. A physical measuring scale is positioned beside the candidate, and a photograph is taken using fixed camera alignment. The captured image is then processed using computer vision to calculate height by referencing the visible scale.

This approach combines Artificial intelligence practices with image-based analysis, allowing height data to be reviewed and stored digitally for later verification.

An automated visual analysis method is applied to interpret image data. Camera placement and distance are maintained uniformly to capture usable images. Convolutional Neural Networks (CNNs) segment each image into identifiable regions, allowing the system to detect key reference points required for height calculation.

Key Functions:

1. Data Collection: Candidate images are collected from examination centers and prepared as datasets for processing within a computer vision system.

2. Annotation: Images are labeled with visible landmarks and scale reference points. These labeled elements are converted into point-based representations and stored as matrix values for calculation.

3. Training: Using deep learning and transfer learning techniques, the model is trained to interpret positional data based on defined X and Y coordinates that determine height values.

4. Testing: The system output is reviewed against expected results. If deviations are found, the model is retrained until acceptable measurement consistency is achieved.

Tools and software used:

Computer vision

Deep learning

CNN

Transfer learning

Features

This AI-based height measurement approach addresses common issues found in manual testing, such as disputes, inconsistency, and missing records. By converting visual data into a verifiable digital format, the process supports documented height assessment and improves trust in physical evaluation workflows.