DEEPCRAFT™ Ready Models
DEEPCRAFT™ Ready Models (formerly Imagimob Ready Models) are production-ready AI / Machine Learning (ML) models that are ready to be added to any edge device. You save all the time, cost and resources required to create collect, validate, and label data, as well as to train custom AI / ML models on that data. We offer a wide range of DEEPCRAFT™ Ready Models developed to meet various business scenarios. Refer to our official website (opens in a new tab) to know more.
How to obtain the DEEPCRAFT™ Ready Models?
-
Navigate to our official website for Ready Models (opens in a new tab).
-
Click the Test Now button of the model you want to download. Fill out the form and submit the form. You will receive an email with a link to download the necessary files and additional information.
-
Open the email and click the Download Your Model link to access the files.
After you have downloaded the required Ready Model refer to the following sections for detailed instructions for testing the Ready Model:
- Flashing DEEPCRAFT™ Ready Models onto the PSoC™ 6 AI Evaluation Kit
- Testing the DEEPCRAFT™ Ready Models on PSOC™ 6 AI Evaluation Kit
- Testing using the Libraries in Custom Applications
Choose Your Ready Model
We offer the following Ready Models:
Audio Ready Models
-
Baby Cry detection - The Baby Cry model is designed to be used in smart home or lifestyle products that want to alert adults to the sound of baby or toddler crying in another room, or to activate baby soothing features. The model can capture most of the baby cry sounds at upto 5 meters distance from device and suitable for 0-4 years old. The model is robust against the most common background sounds, particularly indoor.
-
Cough detection - The Coughing model is designed for use in healthcare products or wearables. It detects coughing, which can indicate illness or other respiratory conditions, and can be used to identify potential environmental hazards. The model can caputure at least 94% of coughs and measures coughs per hour to identify 'sick' versus 'healthy' user. The model is robust against the most common indoor and outdoor background sounds.
-
Direction of Arrival - The Direction of Arrival model is designed to determine the direction from which the sounds are coming from. The uses for this model are many and varied, for example, paired with the Siren Ready Model, it can add safety for drivers, passengers and pedestrians
-
Factory Alarm detection - The Factory Alarms and Sirens model is designed to be used in factory settings to detect alarms and sirens when they go off. This makes factories safer, and can be used for example to alert workers wearing noise cancelling headphones that a siren or alarm has gone off, which they might not otherwise hear it. The model is robust against diverse factory background sounds, and different distances and angels from the sound source.
-
Siren detection - The Siren detection model uses audio event detection to identify emergency vehicle sirens. This can be used to alert pedestrians to emergency vehicles in their vicinity, or to trigger other safety features in wearables. The model can capture over 95% of siren sounds coming from all directions. The model is robust against common traffic background sounds in different environments.
-
Snore detection - The Snore detection model is ideal for companies with healthcare or wearable products that want to identify snoring. This important feature can be used to identify health or environmental issues. The model can capture at least 96% of snores and flexibly designed to be used in wearables or in devices that are placed near the bed. The model is robust against the most common background sounds, particularly indoor.
Watch the YouTube videos to learn how to quickly deploy and evaluate the DEEPCRAFT™ Ready Model for Baby Cry Detection (opens in a new tab), Cough Detection (opens in a new tab), and Siren Detection (opens in a new tab) on a test kit. Simillarly, you can deploy and evaluate other DEEPCRAFT™ Ready Models.
Gesture detection Ready Models
The Gesture Detection model is created to make smart home or lifestyle products easier to use. Wouldn't it be convenien to add time to your microwave by just swiping left or swiping up to turn on your coffee machine. This radar-based model recognizes five hand gestures. The use cases for your products are nearly limitless. The model is robust against many variations, such as sitting or standing, or different user heights.
Fall detection Ready Models
The Fall Detection model is perfect to add into wrist worn devices. It is specifically developed to understand motion and recognize falling, and at the same time reduce battery consumption to a minimum, so that the wearable maintains a long battery life. This model provides safety features for all users, but is particularly useful for seniors.