Taxonomy Icon

Artificial Intelligence

The IBM Model Asset Exchange is a one-stop place for developers to find and use free and open source deep-learning models. Since its first release in early 2018, we have enabled many data scientists and AI developers to easily discover, rate, train, and deploy machine learning and deep-learning models in their AI applications. To continue this effort, we are pleased to announce our third batch of model assets along with code patterns to help you with some handy examples.

flow

These new model assets cover ML/DL domains, including images, audio, text, and time series:

  • Facial Age Estimator – This model estimates biological ages of people given an image. The model first detects and locates faces in the image and then generates a predicted age for each face. The output is the estimated ages, together with the location in the image, of the detected faces. Possible use case: Automating analysis of visitor traffic patterns within a shopping mall, by various demographics, including age.
  • Facial Recognizer – This model recognizes faces in an image and automatically extracts a set of low-level features that represent each face in the form of an embedding vector. These embeddings can be used as input for other machine learning tasks, including classification, clustering, verification, or similarity search. Possible use case: Performing automated security verification against a database of faces, from images and videos
  • Weather Forecaster – This model forecasts near-term weather variables (such as temperature, pressure, or windspeed), given recent historical data. Possible use case: Predicting local weather conditions to enable more focused and efficient targeting of retail offers to customers
  • Name Generator – This model can be trained on a list of names. Once trained, this model can score and suggest names based on the dataset it is trained on. As an example, we have included the Kaggle US Baby Names dataset to train the model on. Possible use case: Generating names and finding the country of origin of a name
  • Audio Sample Generator – This model generates new audio clips that are similar to existing audio clips that the model was trained on. The model can generate short samples of six speech commands (up, down, left, right, stop, go), as well as lo-fi instrumental music. Possible use case: Generating data to increase the robustness of NLP and audio models; increasing data privacy
  • Named Entity Tagger – This model annotates each word or term in a piece of text with a tag representing the entity type, taken from a list of 17 entity tags. Possible use case: Automatically extracting entity tags for use as additional text metadata for data indexing purposes, or as additional features for other NLP tasks

In addition to the new model assets, we also enhanced the deployment and consumption of Model Asset Exchange models:

  • Made all pre-trained models more easily deployable by publishing pre-built DockerHub images.
  • Made all pre-trained models deployable on a Kubernetes cluster.
  • Established public long-running instances for API endpoints of all deployable models and two demo web applications for Image Caption Generator and Object Detector models.
  • Released a new code pattern to show how to create a web app using the Object Detector model.
  • Released a new code pattern to demonstrate how to create a web app using the Facial Age Estimator.
  • Released a new code pattern to show using an image segmentation deep learning model to detect different types of objects from within submitted images.
  • Released an updated version of our Image Caption Generator code pattern.
  • Re-coded the Model Asset Exchange home page and created a new dynamic archive view of models with filter and search capability. Each model also has its own landing page with details that guide the users through the model.

Visit the IBM Model Asset Exchange and browse through these models and enhancements. We hope you can find something that is right for your AI development use cases. We also welcome your comments and suggestions that help us improve and better serve the ML/DL community.