HOOBOX develops high-precision face recognition software for detecting human behaviors. The startup is the one behind 'The Wheelie', the first ever software capable of translating facial expressions, like kisses and smiles into commands to control a wheelchair.

What makes HOOBOX's face recognition solution unique is that our algorithms were developed specialized for healthcare applications, being able to capture facial expressions in realtime and analyze more than 90 key facial points with high precision regardless of light condition or head position.

The Wheelie is part of the next generation of facial recognition software specialized in capture human behaviors using A.I. Running in the background of any application, the Wheelie can monitor patients, capturing spasms, agitation and sedation levels, and 10 level of pain.

HOOBOX Robotics was founded in 2016, in São Paulo, Brazil. One year later, the company started its operation in Houston at JLABS@TMC, the Johnson & Johnson Innovation Lab.

Key Links


Wheelchair users who lost partially or completely the movement of arms and hands cannot control a joystick of a motorized wheelchair in order to have their mobility and autonomy back. Only in the U.S., we have 1.2M people with spinal cord injury, stroke, brain injury, amyotrophic lateral sclerosis (ALS). In Brazil and U.K, around 800,000 people.

There is a wide number of handsfree mechanisms addressing this problem, but the challenge raised by the users is that these technologies require them to wear sensors physically attached to the body which tends to be awkward and uncomfortable.

To address to this challenge, we developed a computer vision-based solution called 'The Wheelie', the first ever computer program capable of translating facial expressions, like a kiss, a smile into commands to control a wheelchair. With the Wheelie, people with upper limb activity limitation can control their own wheelchair using facial expressions captured only by our A.I. software and a camera mounted on the wheelchair.

To make our application even more scalable, we are using Intel® Optimization for TensorFlow along with the Intel team extensive support on the inference time optimization. As hardware, Xeon Scalable Processors has been used to train the machine learning algorithms.