Automated Detection and Segmentation of Mechanical, Electrical, and Plumbing Components in Indoor Environments by Using the YOLACT++ Architecture
Bibliographic record
Abstract
Indoor construction environments, with their high density and detailed components, are complex areas for progress monitoring and reporting. Traditional manual monitoring systems, often constrained by poor lighting and accessibility, are inaccurate and time-consuming. With recent technological advancements, deep learning-based object recognition models have achieved considerable attention in construction. This paper introduces a novel method for progress monitoring and reporting of construction operations, employing digital imaging and the You Only Look At CoefficienTs (YOLACT++) deep learning algorithm to automatically recognize mechanical, electrical, and plumbing (MEP) components in challenging indoor settings. Data augmentation techniques and transfer learning were applied to improve the model’s generalization and adaptability. The study distinctively focuses on complex components in complicated indoor environments, a less explored area in current research that mainly centered on outdoor or simpler indoor settings. To achieve this, the study enhanced the dataset quality by generating synthetic images that closely represent actual indoor conditions including different lighting, object complexity and scale, occlusion, clutter, and viewpoints. This study also evaluated different mixes of synthetic and real images to determine the optimum combination for effective training. Moving beyond commonly used algorithms such as Mask R-CNN and You Only Look Once (YOLO), the method applied in this work is the YOLACT++ with deformable convolutional neural networks v2 (DCNv2), enhancing the model’s ability to handle objects with different scales, postures, rotations, and viewpoints in the images that are essential in indoor environments. The model is validated on a large test dataset, including real images from construction sites, to cover different indoor scenarios. The model achieved a precision of 84.80% and a recall of 85.58% for HVAC duct detection and a precision of 86.87% and a recall of 73.93% for pipe detection, demonstrating its effectiveness under challenging conditions. This method contributes to more accurate automated progress monitoring in indoor environments by reducing manual and error prone inspections.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.000 | 0.000 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.001 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.002 | 0.002 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".