The Challenge
- High distortion of images
- Poor training data in terms of quality and quantity
The Approach
- Generation of high-quality augmented training data using customized method for generation of fisheye images from normal images.
- Development of deep learning architectures optimized specifically for fish-eye lens images.
- Ensemble methods combining multiple models for improved performance
Result & Added Value
- The architecture not only could handle the large distortion of fisheye images but also could deal with the lack of data.
Discover more
Related Projects
Technology Consulting
Our Competencies
With INVENSITY by your side, you have access to our wide scope of competencies, in-depth knowledge, and our highly skilled and experienced team of consultants.
