Computer vision for site safety and housekeeping

Using computer vision technologies to enhance the safety of construction sites by automatically detect hazardous activities and poor housekeeping, and alert site personnel.

cctv-48

Housekeeping detection

cctv-48

Change detection

cctv-48

Open edges and workers under load

Traditional site and safety management rely heavily on manpower

Frequent safety violations leading to costly fines

Difficult in maintaining consistent safety monitoring across large sites

Manual safety inspections are inefficient and prone to human error

Our approach and solutions

cv_framework

Poor housekeeping detection

Improving the overall safety standards, safety climate, and productivity through detecting poor housekeeping hazards on-site

Our solution can detect the objects:

  1. Rebars
  2. Pipes
  3. Water ponding
cv_housekeeping

Actual results from our object detection solution

Rebars

Pipes

Water ponding

Change detection

Change Detection for Housekeeping Dataset

Codes and dataset for housekeeping change detection network using feature fusion and large vision models.

changedet

Samples from our change detection for housekeeping dataset

cd1
cd2
cd3

Open edges and workers under load

Developed an object detection model based on TensorFlow platform and applied image processing techniques like OpenCV library to detect worksite hazards like working on open edges and workers under load

Our solution can detect:

  1. Workers under load
  2. Workers near the open edges
cv_01