Early detection of leakages

In buildings, leakages are noticed late, either due to furniture blocking the view or that portion not visible from outside. Build algorithm which will capture “images” and classify them as “leakage / no leakage”. This can be further enhanced by using selfie sticks / drones, for not so easily accessible areas.

Project State

Public Project

Licences

Software Licence: Project has no software
Hardware Licence: Project has no hardware

Admins

banawalikar

Members

Does this project pique your interest?

Login or register to join or follow this project.

Reason for subject selection :

https://www.linkedin.com/pulse/water-leaks-multi-storey-buildings-problem-bigger-than-hamid-khan

Water leaks are exceedingly detrimental to the health of high rise buildings. Water could enter the building envelope through different paths – from damp construction materials during the construction stage, through leaking roofs, basements, water features, wet areas and leaking water installations. Concrete being a permeable material, leaking water will find its way in and spread easily. If the source of the water leakage is left unattended, it can cause significant structural damage that often needs very expensive rectification to structural elements.

Water leak is the major cause of early onset of corrosion and concrete deterioration.

In Hong Kong, the Department of Health identified water seepages in buildings as the possible source of the outbreak of the severe acute respiratory syndrome (SARS) epidemic in 2003. Water leakages in the buildings due to poor maintenance and building defects could lead to growth of toxic mold that could cause serious health and safety issues for the occupants. There are also established evidences of the relationship between water leaks in buildings and respiratory symptoms in occupants.

  1. Was thinking of a possible “preventive” solution for a very long time.
  2. The “smart home contest” with starter kit, helped in thinking a probable solution.

Methodology :

  1. Would have collected a database of "leakage / no leakages" snaps for model to train.
  2. Once the model was "trained" and "tested", would have deployed to classify, a snap as "leakage" or "no leakage".
  3. On basis of actual observations, would have tried to improve "accuracy" of the model.

Challenges :

  1. Would like the model to be very accurate. Ok with “false positives” however definitely would like the “false negatives” to be as minimum as possible. Due to this, would have preferred a very large “training database” of leakage snaps. Not able to arrange for the same.

Next Steps :

If this "preliminary" step, would have been successful, then ..

  1. Use with remote controlled drones, to access, not so easy to access areas such as higher floors / ducts
  2. Modify it to be used with “selfie sticks” to access, not so easy to access areas such as behind furniture etc
  3. Privacy : No need of a stranger accessing the apartment to check for leakages. With a hand controlled camera (?), the tenant can capture photos of the possible affected areas and pass it on, to be analyzed.
Name Article number Link Quantity Unit Price
UP Embedded Starter Vision Kit 1 349.00 $
Wireless Mouse / Keyboard 1 15.00 $
HDMI Type A Male Cable 1 5.00 $
TV Monitor 1 70.00 $
Total 439.00 $
Task Owner Creation Date
Will try to finish the project
banawalikar
2018-11-30

Comments

Ready to join the project?

You'd like to participate ... Show more