Sponsored by Infineon, the Getting Edgy with Machine Studying contest showcased modern edge machine studying functions.
Members had been challenged to make use of DEEPCRAFT Studio and the PSOC 6 AI Equipment in creating ML initiatives for good residence and industrial functions. They needed to construct an edge AI mannequin by accumulating sensor information leveraging the equipment’s onboard sensors, showcase the mannequin working in the true world, and doc the answer as a Hackster tutorial.
After reviewing all of the submissions, listed here are the eight initiatives that emerged as winners.
AI Detection of Cardiac Abnormalities
The guts is an important muscle within the physique, and coronary heart dysfunction is a number one reason for loss of life, however for varied causes, not everybody retains it in nice situation. Extra individuals want common cardiac exams to mitigate coronary heart dangers. Testing typically requires cumbersome gear in costly hospitals. So, EdOliver and Victor Altamirano developed a wearable electrocardiograph utilizing dry electrodes and SparkFun’s AD8232 coronary heart price monitor.
The dry electrodes make the electrocardiograph usable by athletes throughout exercise. Utilizing machine studying, the collected information is organized and offered to carers in a approach that interprets to actionable insights, addressing the issue of knowledge overload from wearable well being units.
Do understand that this venture is for “demonstrative and illustrative functions solely” and shouldn’t be used instead of an authorized medical machine.
AI-Powered Predictive Industrial Machine Upkeep
Predictive upkeep employs information and analytics to mitigate the danger of apparatus breakdown and failure, improve gear lifespan, and scale back the necessity for alternative machines.
Eslam Fayed’s venture is a predictive upkeep mannequin for industrial machines. It makes use of the to gather real-time vibration information. The mannequin’s coaching information was collected utilizing PSOC 6’s built-in inertial measurement unit (IMU), educated in DEEPCRAFT Studio, and deployed to the PSOC 6 by way of ModusToolbox.
Fayed exams his mannequin on a vacuum cleaner, however the venture will be tailored to different industrial machines.
SmartListener: Ambient Sound Classifier
Sound classification makes use of waveform traits to group audio alerts into predefined classes for functions like voice recognition, music style detection, and safety monitoring.
Mohammed Ali Bedair created the SmartListener to categorise sounds in a house surroundings. The mannequin was educated utilizing audio samples from the ESC-50 (environmental sound classification) dataset, transformed to a 16kHz bit price for compatibility with DEEPCRAFT Studio.
There are 5 separate lessons for sound classification, together with child crying, fireplace alarm, glass breaking, footsteps, and barks. The machine listens for these particular sounds and sends an MQTT alert when they’re detected. It makes use of a TensorFlow Lite mannequin educated in DEEPCRAFT Studio and has a 3D-printed enclosure for deployment in any room.
SmartScale: Produce Freshness Detection
SmartScale {hardware} structure
Day-after-day, 1000’s of completely edible fruit and veggies are disposed of as a result of they fail to fulfill beauty requirements. Image-perfect produce is an unrealistic commonplace that results in giant quantities of meals waste. Some supermarkets are working to scale back this waste by donating surplus meals and promoting “ugly” fruit and veggies at a reduction.
Milan Ferus-Comelo’s SmartScale goals to introduce an goal commonplace for “evaluating and promoting contemporary produce.” Since look doesn’t inform the complete story, it makes use of radar imaging and AI to “assess the true inside and exterior situation of fruit and veggies.”
It really works as a part of a retail checkout expertise the place a buyer locations a fruit merchandise on the SmartScale, and the machine calculates a freshness rating utilizing the built-in neural community. The SmartScale applies a proportional low cost based mostly on the rating and prints a value label for the shopper. It reduces waste and allows prospects to make sustainable selections.
It presently works with bananas, however will be prolonged to extra fruit and veggies.
THEIA: Monitoring Heuristic Analysis Intelligence Analyzer
Situational consciousness is an important a part of residence safety and automation. Bryan Staley and Brayden’s venture makes use of the PSOC 6’s radar sensor and an edge AI mannequin to analyze doorway site visitors and classify passing objects.
The venture contains three separate functions for coaching, inference, and processing: THEIA-REC data and downsamples coaching information, THEIA-INF runs mannequin inference, and THEIA-CLIENT processes the outcomes and shows the knowledge to the person.
As a result of time constraints, the mannequin has two major lessons: individual and ball. It could possibly detect an individual and a ball shifting out and in of the radar sensor’s discipline of view.
AI Blender Velocity Detector
Guillermo Perez Guillen’s venture detects the present standing of a blender utilizing the PSOC 6’s onboard accelerometer and provides a corresponding voice or LED alert. to detect when the blender is in operation. The thought will be utilized to any machine that operates with DC motors in any setting.
The board needed to be positioned in numerous orientations to seize at the very least two axes of the accelerometer. The venture helps LEDs and voice alerts by way of the EDU DFR0699 voice recorder module.
Unlawful Logging Detector
Unlawful logging is a serious concern in most growing nations. The Congo Basin, for instance, loses 1 to five% of its forest cowl yearly because of unlawful logging, and a big proportion of wooden consumed in Mexico is of unlawful origin, based on a 2025 research.
Alejandro Sanchez’s machine detects chainsaw sounds and human voices in protected areas, signaling that unlawful logging could also be occurring. The PSOC 6’s onboard microphone obtains an audio sign, which the mannequin classifies into certainly one of three lessons: chainsawI, individuals, or forest (base state).
IntelliFan
There may be nothing fairly like a cool breeze blowing within the warmth of summer season. Most followers blow in a single course and wish handbook adjustment. Wouldn’t or not it’s cool in case your fan adopted you across the room fairly than the best way round? That’s why Ritik and Zilu of De Anza Faculty constructed IntelliFan, a fan that tracks individuals mechanically and responds at hand gestures.
The venture makes use of DEEPCRAFT’s pretrained gesture mannequin and PSOC 6’s radar sensor to acknowledge 5 distinct swipe and push gestures. The net interface shows a dwell digital camera view and has controls for pace, monitoring, and handbook positioning. The staff plans to improve the present product with voice instructions, a battery for transportable use, and a tower design for higher airflow.
