Image
Information-based waste collection
Published on 27 November 2019
Netherlands
This is the good practice's implementation level. It can be national, regional or local.
About this good practice
Amsterdam copes with a difficult waste system. Roughly 90 % of the residual waste is collected by underground waste containers, of which Amsterdam has 13.000. Although the system is efficient when it comes to logistics, the system also comes with great littering. A big city as Amsterdam is anonymous and diverse, which unfortunately makes the waste containers an easy target for pollution. Object detection is applied in this context.
Object detection is to be used to (a) fast and just recognition and response on littering (around waste containers) and (b) to optimize the logistics for operational services involved in collecting waste (waste department and law enforcement, i.e.).
Amsterdam is developing and testing the mechanism of object detection where:
- Images of all appliances are collected;
- Images are labelled with characteristics (e.g. “garbage bag”)
- Labelled images are classified and transferred to working orders, which in turn are divided and sent to the right execution service (e.g. “garbage route 1”, “garbage route 2”, “law enforcer 1” etc.)
How does the practice reach its objectives and how it is implemented?
It is in an experimenting phase.
Who are the main stakeholders and beneficiaries of the practice?
The public services involved in maintaining the public space, such as waste services, law enforcement and street sanitary service. The system is developed by CTO (chief technology office) Office Amsterdam and the universities of Amsterdam.
Object detection is to be used to (a) fast and just recognition and response on littering (around waste containers) and (b) to optimize the logistics for operational services involved in collecting waste (waste department and law enforcement, i.e.).
Amsterdam is developing and testing the mechanism of object detection where:
- Images of all appliances are collected;
- Images are labelled with characteristics (e.g. “garbage bag”)
- Labelled images are classified and transferred to working orders, which in turn are divided and sent to the right execution service (e.g. “garbage route 1”, “garbage route 2”, “law enforcer 1” etc.)
How does the practice reach its objectives and how it is implemented?
It is in an experimenting phase.
Who are the main stakeholders and beneficiaries of the practice?
The public services involved in maintaining the public space, such as waste services, law enforcement and street sanitary service. The system is developed by CTO (chief technology office) Office Amsterdam and the universities of Amsterdam.
Resources needed
- Data (Images and videos);
- Model development;.
- Hardware (cameras, servers);
- For the first experiment 4-6 months the expenses are estimated on € 20.000.
- Model development;.
- Hardware (cameras, servers);
- For the first experiment 4-6 months the expenses are estimated on € 20.000.
Evidence of success
Still in experiment phase.
Potential for learning or transfer
Experiment experiences, success and fail factors can be shared;
Model structure;
Costs of software and model development;
Costs of hardware.
Model structure;
Costs of software and model development;
Costs of hardware.
Good practice owner
You can contact the good practice owner below for more detailed information.
Organisation
Municipality of Amsterdam
Netherlands
Noord-Holland
Contact
Communications manager