Projects Proposed Projects
Please see propose projects for Spring and Winter upcoming semesters. This is a 4 academic point course. For further details please contact the laboratory engineer Yaron Honen (04-8295535, room 441).
A video-based detector for attention-slips in children
This project is a collaboration between Geometric Image Processing Lab (GIP) and Educational Neuroimaging Center (ENC).
An innovative intervention program of joint storytelling was used to promote cognitive abilities, sustained attention and language abilities in 4-6 year old children. Videos of these children during the intervention sessions were recorded using a web camera. We aim to quantify the children’s movement levels along the intervention sessions in order to explore an early bio-marker for ADHD and its relation to school readiness.
The project’s main objectives are -
1. Detect children from the video and follow their movements.
2. Operate a motion parameter to receive a movement score for each child over each intervention session.
3. Compare the motion parameter with different behavioral tests scores preformed with this cohort of children.
• Courses: one of the following - Computer Vision, Image Processing, Excellence students
Emoji-kid recognition from multiple web cameras videos
1. Synchronize the different web cameras videos in order to detect best view of each child.
2. Classify each child emotion state into 3-4 of emotional categories as a function of time.
3. Output a video of the recorded session where the faces of the children are replaced with a representing emoji
• Courses:one of the followings - Computer Vision, Image Processing, Computer Graphics
Drone Navigation using 3D sensing
The drone, packed with a 3d sensing camera and a powerful on-board computer, is aimed to achieve autonomous flight capabilities
Yaron Honen: room 441 Taub building
. Seeking for bright and talented students for Technion Drone Project.
Students in this project must already obtain:
- Excellent knowledge in c++
- Good knowledge in OOP
- Good knowloage in Linux echosystem
Real-time dense facial motion capture
Project goal: to track the geometric structure of the face from a webcam in real-time.
Recently, we demonstrated that it is possible to recover the geometric structure of the face and to track it over time with the help of deep convolutional network.
In this project, we would like to construct a real time framework for tracking the dense geometric structure of the face across frames.
The students will learn to deploy convolutional networks for real time applications, geometric processing algorithms as well as rendering
- Excellence students
- Programming experience: C++ and Python.
- Courses: at least one of the following courses - Computer Vision, Image Processing, Computer Graphics or Geometric Computer Vision.
Embody game
Use your body silhouette to help the hero, by using RealSense
R-Password a 3D password
A new approach for password in the AR/VR domains
Use RealSense to generate a 3d password by drawing a 3d scribble
Sight beyond sight
A bucket of tools which enhancements the human vision sight, for example ‘lighting’ a dark.
Explore human emotion
Photorealistic Augmented Reality
Embed a 3d model into the real environment by learning the illumination model apply shading (if needed) to have a realistic augmentation
Real-time photorealistic Object rendering
In this project we will implement a fast rendering (as the web page loads) of a user provided image mapped to a user selected item (mug, pillow, candle, etc.), intended for a gift or souvenir.
Make your family album
The goal of this project is to fuse family pictures into famous works of art. Light modeling, geometric transfomations, image segmentation, image registration are the tools of the trade.
Augmented reality in road navigation
The purpose of this project is to create augmented reality on a mobile phone for road navigation. The project is composed of several parts from which the students can choose to focus on: Simultaneous localization and mapping, Augmented reality, road detection and mobile programming.
Gil Shamai: room 443
good programming skills.
UI/UX Augmented widgets
Design & builds an exciting UX/UI widgets:
1. Placing the GUI in an optimal distance
2. Adding gravitational rules to the GUI
3. Adding “Negative shadow”*(Soft glow)
4. Stationary Vs. animation
5. Avoid hand(Virtual) tremor
Guliver - Realtime head positioning and tracking
Gulliver is the next step in the evolution of virtual reality. Seamlessly transitioning a person into a 3D reconstructed digital world, using a kit composed of an Oculus Rift, an Intel RealSense depth camera and a computer.First, the Oculus Rift allows for the “basic” virtual reality experience, which lacks the ability to receive input simply by having the user move around and interact with objects.Building upon this, the user is positioned inside the 3D digital world. We are then able to receive his movements as our input. This allows for a more natural interaction with the environment – moving through the world simply by walking, interacting with objects with your own hands. No conventional input device needed.Finally, we make use of a depth camera. Positioned on the front of the Oculus Rift and acting as the user's eyes in the real world.This allows for the projection of the user's hands from the real world, into the 3D digital world, making the entire experience a lot more believable.Imagine walking through the Lego castle you've built as a child. Raising your hands in the air and watching them appear, just as they would in the real world, in front of you. Walking alongside the other Lego people in the castle.
UNITY-based Simulation of Intersection with no Traffic Signs and no Traffic Lights
סימולציה של כלי רכב שונים והולכי רגל החוצים צומת (ללא התנגשויות) בשימוש אך ורק ב Visual looming שאינו דורש שחזור "תלת מימד". ישקל שימוש בטכניקת לימוד עמוק (deep learning ) לשם טיפול ברעשים ועיוותי תמונה.
Obstacle Avoidance in Virtual Reality Using the Visual Looming Effect
הולך רגל עם משקפי Virtual Reality נמנע מהתנגשויות ע"י מיפוי של סביבה רלוונטית בשימוש אך ורק ב Visual looming
“Ant-Like” Multi Robot Tracking Using Relative Changes in Visual Information
דימוי בעזרת רובוטים של התנהגות יצורים דמויי נמלים. כל רובוט עוקב בעזרת מצלמה/מצלמות אחרי הרובוט שלפניו. הרעיון הוא להשתמש ב- Visual looming . ישקל שימוש בטכניקת לימוד עמוק (deep learning ) לשם טיפול ברעשים ועיוותי תמונה.
.SWARM Behavior Using Relative Changes in Visual Information
דימוי בעזרת רחפנים (או רובוטים) של התנהגות ציפורים העפות בלהקה.
כל רחפן עוקב בעזרת מצלמה/מצלמות אחרי רחפנים בסביבתו המיידית.
הרעיון הוא להשתמש ב Visual looming שאינו דורש שחזור "תלת מימד"
ישקל שימוש בטכניקת לימוד עמוק (deep learning ) לשם טיפול ברעשים ועיוותי תמונה.
Keeping Constant Distance from a Wall Using Image Sequence
שמירת מרחק קבוע ממשטח (תוך כדי תנועה במקביל אליו) ע"י שמירת פוקוס קבוע בשימוש באלגוריתם המבוסס על אפקט Visual Threat Cue שעושה שימוש בשינוי היחסי בפוקוס בין תמונות עוקבות + בלי צורך בשחזור "תלת מימד". ישקל שימוש בטכניקת לימוד עמוק (deep learning ) לשם טיפול ברעשים ועיוותי תמונה.
Visual Navigation
ניווט רובוטים או רחפנים בסביבה לא מוכרת
ודינמית בשימוש באלגוריתם המבוסס על אפקט
Visual Looming שעושה שימוש בשינוי היחסי
בין תמונות עוקבות למטרת ניווט בין מכשולים + בלי צורך בשחזור "תלת
מימד". ישקל שימוש בטכניקת לימוד
עמוק (deep
learning ) לשם טיפול
ברעשים ועיוותי תמונה.
Historical map allignment
In this project we will implement a matching algorithm between historical maps and modern ones (google world). Several allignment methods will be tested and the maps will be overlayed.
Detection of violence in nursing homes
Surveillance cameras are already there. In this project we will automatically detect violent behaviour of staff towards the elderly and helpless.
Analysis of teachers movements in classroom
Using videos of apprentice teachers in classroom, we will inspect their body movements and face orientation. Quantitative and qualitative results will help them perform better in class.
Deep learning automatic recognition of license plates of vehicles
This project will implement a system to automatically read license plates of the cars (LPR - LICENSE PLATE RECOGNITION). The system will receive input images of the cars provided different location and lighting conditions, extracting the image from the location of the license plates of vehicles, and decipher the digits of each vehicle. In this project we will use segmentation algorithms and deep neural/convolution network classification.
Autonomous Outer Body Perception
Build an indoor autonomous drone/s with a given indoor blueprint that gathers and record video and information. (Light condition, visibility, obstacles…)
Rubik’s Solver
In this project we will implement an assistant solver for the rubik’s cube. We will use a virtual reality headset and a camera to scan the cube and overlay the next move directly on the cube, from the user’s viewpoint
Augmented Graffiti
In this project we will implement an augmented reality graffiti applications.
Similarly to the deceased Stiktu application, users of the app will first take a picture from their surrounding and then draw some graffiti on it. User arriving at the same location can then view the graffiti by looking at the graffiti through their mobile phone
Mixed reality games
Choose a game (Chess, backgammon, checkers, Pokémon…) and suggests the next moves on the board. Recording (with time-stamp a session will be nice edit value. It means that a user can load game and can step to a specific time in the game and try different approach
Mixed reality – Stereo immersive augmented reality system
Build ecosystem of immersive augmented into Oculus Rift/HTC vive with a pair of calibrated cameras. It includes building a stereo camera rig. (Can use our 3d printer)
3D sculpture editor
Building an exciting user experience for artist's modelling programs. (Like Pixologic – Sculptris application). The sculpture language should include: volume, surface, material, texture, transparency, mass and impermeability.
Computer Vision Based Bike Fit
In this project we will implement a system that can find the right parameters for adapting the bicycle to the rider. We will use 3D or 2D camera to capture the scene and than analyze the angles and distances.
Ron Kimmel
Eye tracjing for AR/VR
Build a stereo camera system customized specifically for the Oculus and perform eye-tracking in real time. Build a demo that enables a) user interaction using eye-pointing b) depth of focus for individual objects
- Projects:
- Proposed Projects
- Last Projects
