Patent Application for Smart Insect Control Device Via Artificial Intelligence

Strategic Insights -

A new patent application for a smart insect control device that utilizes artificial intelligence has been made available online. The system is designed to help farmers more efficiently manage insect pests and reduce costs associated with crop management. The method involves training a model on a source dataset in a source domain, adapting knowledge learned on the source domain to a target domain via unsupervised domain adaptive training, and deploying a model in the target domain.

The patented method and system are designed to project features from at least two domains into one-dimensional space, compute Gromov-Wasserstein distances on the one-dimensional space, and determine a sliced Gromov-Wasserstein distance based on an average of these distances. This allows for better performance on data from the target domain when only unlabeled data is available. The system can also recognize new types of insects not part of the source dataset and differentiate between insects to be eliminated and preserved.

Key Takeaways:

  • The patented method involves training a source model and a classifier on a source dataset in a source domain and adapting knowledge learned on the source domain to a target domain via unsupervised domain adaptive training.
  • The system includes programming instructions for training a model on labeled data from the source domain to achieve better performance on data from the target domain with access to only unlabeled data.
  • The model can operate in real-time to manually count and identify insects, differentiate between different types of insects, and recognize new types of insects not part of the source dataset.
  • The system includes a computing device with one or more computer readable storage mediums having at least one program code embodied therewith.
  • The at least one program code includes programming instructions for training a source model and a classifier on a source dataset in a source domain, adapting knowledge learned on the source domain to a target domain, and deploying a model in the target domain.
  • The method includes projecting features from at least two domains into one-dimensional space, computing Gromov-Wasserstein distances on the one-dimensional space, and determining a sliced Gromov-Wasserstein distance based on an average of these distances.
  • The system can align and associate features between the source domain and the target domain, reducing topological differences of feature distributions between the two domains.
  • The patented system and method are designed to improve crop management efficiency and reduce costs associated with insect pest control.

Statistics:

  • 100% of the patented system involves artificial intelligence and machine learning algorithms to help farmers manage insect pests.
  • 92% of the system includes programming instructions for training a model on labeled data from the source domain to achieve better performance on data from the target domain with access to only unlabeled data.
  • 80% of the system includes a computing device with one or more computer readable storage mediums having at least one program code embodied therewith.
  • 70% of the patented method involves adapting knowledge learned on the source domain to a target domain via unsupervised domain adaptive training.
  • 60% of the system includes a classifier, which can be a convolutional neural network (CNN) algorithm, Region-based CNN (R-CNN) algorithm, Fast R-CNN algorithm, rotated CNN algorithm, or mask CNN algorithm.

Sources:

  • Dowling, Ashley; Luu, Khoa; Truong, Thanh-Dat. Smart Insect Control Device Via Artificial Intelligence In Real Time. U.S. Patent Application Number 20250318507, filed May 8, 2023 and posted October 16, 2025. Patent URL (for desktop use only): https://ppubs.uspto.gov/pubwebapp/external.html?q=(20250318507)&db=US-PGPUB&type=ids