Computer Vision Engineer
To apply, send resume and if applicable, a portfolio of your work to
1. Work with both 2D (RGB) and 3D (Depth) raw perception data, sometimes even with other sensors who provides other modalities of percepts of the world.
2. Solving vision tasks include but not limited to:
Vision sensor calibration (internal and external)
Identify and classify objects, generate bounding boxes, segment objects/ generate object masks
Measure object size in physical unit
Estimate and tracking object pose
Learn/ scan new objects in short time, possibly with surface reconstruction
Identify and read barcodes
Large scale automatic data collection
Efficient (even autonomous) data labeling
Vision based mapping, localization and navigation
3. Perform these tasks (if applicable) efficiently and robustly, against various lighting conditions/ under heavy occlusion/ while the relative pose between camera and object are changing as high speed/ multiple objects at the same time/ multiple cameras at the same time/ across different sensors/ deploy them on real robot hardware etc.
4. Work on challenging scenarios such as partially observable, sequential, dynamic, continuous, and unstructured environment.
5. The job might involve researching on particular topic, reading papers, attending conferences, efficiently implementing algorithms on hardware, testing in real environment, debugging, parallelization, sourcing sensors and computing platforms, writing documents/ papers, generating patents, and other possible forms.
1. Background in computer vision, computer graphics, image processing, machine learning or other related fields
2. Proficient with C++ and Python (knowledge and daily experience)
3. Solid knowledge in computer vision theory with past experience of turning those theories into robust program.
4. Proficient with vision library: OpenCV, PCL, etc.
5. Proficient with deep learning frameworks: Tensor Flow, Keras, etc.
6. Skilled with machine learning techniques, especially deep learning, reinforcement learning and transfer learning.
1. Have publications in top Computer Vision and Graphics journals and conferences such as CVPR, ICCV, SIGGRAPH etc.
2. Experience in programming, robotic or vision algorithm challenges, e.x. ILSVRC
3. Strong experience in research/ engineering in any of the following areas:
Image detection and segmentation
Multi-object pose estimation
Have created and maintained large-scale datasets (> 10000)
Multi-cluster GPU training
Solved difficult vision problems in industry
published interesting code on github and/or maintains an active blog on vision & learning
created vision system by extensively building upon previous open-source work to solve a unique problem
have created unique dataset