FB Pixel

BoxDetection – A Warehouse Anomaly Detection Solution


Innovator partner’s Representative nameHuynh Ngoc Tan

Solution domain: Machine Learning

What is the Problem that this Solution contributes to solve?

Detect misplaced box in warehouses via camera. The solution tries to solve the problem of abnormal detection in the warehouse (dropped cargo detecting to be more specific) using various AI techniques & areas like object detection, lane detection, segmentation, abnormal detection.

At the PoC Development Stage, this solution is not good enough yet to detect the abnormal in the warehouse or detect dropped box, but can be utilized in several cases like counting boxes and detect strangers in the warehouse

What are the Added Values that this Solution brings to the Problem?

This solution introduces one of the approaches to the abnormal detection problem in the warehouse so that the licensees can have an initial assess about the effectiveness or potential of the approach.

Who are the potential Licensees that this Solution can be helpful for?

  • The warehouse owners or stock owners who need to detect misplaced box via camera
  • Researchers

In what ways can this Solution be used by those Licensees? Any advice for the most effective application (that the Licensees can make the most of this Solution)

The solution can be used in the warehouse, stock to count the number of boxes on the shelves, or to detect how much space left on the shelves, or percentage of the shelves space has been used. Potentially can be a starting point for detecting abnormal in the warehouse.

The next development for this Solution that the Licensor can continue to work on (or the Licensees may consider).

  • Develop an application to help counting the box, calculating the occupied shelves spaces.
  • Extend the model to detect more types of cargos with various shapes.
  • For the application of dropped cargo detection in the warehouses, the accuracy should be improved and tested in various warehouses.

Classifed based on Kambria NFT IP Pricing Framework (guidelines):

  • Solution Development Stage: PoC
  • Level: 1

License Price: 1,000 USD

Number of Licenses: 50

  • Each license is an NFT with the solution packaged in the private unlockable content.
  • Solution package includes: Introduction deck, Demo video, Sourcecode and Technical document.