Delivery date: 10 September 2009
Description: Architecture, development and implementation in Python with wxWidgets and OpenCV libraries of “interface with graphical editor for automatic labeling of image sequences” for files creation process “Ground truth” needed to evaluate the recognition objects of a mobile robot. In collaboration with the department of robotics and artificial intelligence IIIA (Institut de Investigació de Intel·ligència Artificial at UAB).
The adaptation of object recognition for mobile robotics requires a new approach and applications to optimize training robots to obtain satisfactory results. It is known that the training process is long and tedious, where human intervention is absolutely necessary to monitor the behavior of the robot and the direction toward the goals.
It is for this reason that it has developed a tool that significantly reduces human effort should be made to this monitoring, automating the needed process for an outcome evaluation and minimizing the time wasted due to human error or lack of infrastructure .
The graphical interface is capable of opening sequences of photographs, creating a user bounding boxes (straight lines) on certain objects as they appear in the image sequence, which is divided into a video of photographs in which objects moves through the pictures in the same direction.
Software performs image processing on previous photographs and to automatically track to locate objects in new images previously defined, implemented using image processing algorithms as “meanshift” and the HSV color model.
For more details, consult the project report:
Computer Engineering Final Year project Etiketagger in pdf format.
Development time: 7 months
Delivery date: 13 May 2009
Link: – link not available –
Description: Fellow at ENDESA NETWORK FACTORY (ENF), under the tutelage of Felipe Alvarez-Cuevas and supervision of Javier Oviedo, R & D in developing embedded software for a remote control on household chores processor: “PROXIMA GSM” project. It is a software project management techniques and automation alarms.
Performed using motion detection devices and floods, wired or via radio (wireless) connected to a device with an embedded operating system (embedded linux kernel 2.6.x). The development was made based on the communication protocol using technical alarms with their respective central processing. Data packets are captured, processed and created by software implemented in C language
PHASE 1 -. Through the software, it was able to sound an alarm and send SMS text messages through a GSM chip installed on the device, when activated a technical alarms.
The project was initiated using the explanatory documentation and data packets based protocol used for communication between devices, central alarm, provided by the company that owns JR Sistemas de Seguridad – Security Systems capturing data packets sent by the alarm, and building new ones for new functionalities, by communication via cable RS-485 and via radio.
For more detail see extra documentation:
Memory ProximaGSM v1.0 in pdf format.
Development time: 8 – 10 months