Development of an Image Data Set of Construction Machines for Deep Learning Object Detection
Why this work is in the frame
A frame that forgets how it found something cannot be audited. These are the routes that admitted this work.
Bibliographic record
Abstract
Deep learning object detection algorithms have proven their capacity to identify a variety of objects from images and videos in near real-time speed. The construction industry can potentially benefit from this machine intelligence by linking algorithms with construction videos to automatically analyze productivity and monitor activities from a safety perspective. However, an effective image data set of construction machines for training deep learning object detection algorithms is not currently available due to the limited accessibility of construction images, the time-and-labor-intensiveness of manual annotations, and the knowledge base required in terms of both construction and deep learning. This research presents a case study on developing an image data set specifically for construction machines named the Alberta Construction Image Data Set (ACID). In the case of ACID, 10,000 images belonging to 10 types of construction machines are manually collected and annotated with machine types and their corresponding positions on the images. To validate the feasibility of ACID, we train the data set using four existing deep learning object detection algorithms, including YOLO-v3, Inception-SSD, R-FCN-ResNet101, and Faster-RCNN-ResNet101. The mean average precision (mAP) is 83.0% for Inception-SSD, 87.8% for YOLO-v3, 88.8% for R-FCN-ResNet101, and 89.2% for Faster-RCNN-ResNet101. The average detection speed of the four algorithms is 16.7 frames per second (fps), which satisfies the needs of most studies in the field of automation in construction.
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.
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.000 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
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 it