The need to incorporate artificial intelligence in real-life scenarios and essentially improve the quality of various industry business services rise the need to develop datasets that fit very specific different needs. In our case, we intend to develop an algorithm that augments an annotated dataset for retail stores. The tricky part is to augment a dataset that forces our detector to be robust and provide object detection with high accuracy under different conditions.
Basic knowledge of computer vision and deep learning. Programming skills: python, TensorFlow (optional), PyTorch (optional).
The desired outcome is a network that is invariant to illumination, scale and rotation of the products on the shelf.
Expected deliverables: Final report, code base, dataset