The Computer Vision Journey is a power packed set of 6 courses that help engineers go beyond basic Computer Vision skills.
The courses inside are:
The Tracking Journey is helping engineers implement a true Perception system that goes beyond simple object detection, and implements object prediction and tracking.
The courses inside are:
The LiDAR Journey is for engineers who wish to understand all about LiDAR, from basic Point Clouds Processing, to advanced 3D Object Detection.
The Robotics Journey is here to help engineers understand self-driving car and robot architectures, and export their own projects in robotics platforms.
The Advanced Deep Learning Journey teaches advanced Deep Learning and Research concepts to optimize neural networks and ship them to production environments.
For example, here is one of my courses from the Tracking Journey, teaching 4D Perception and advancec concepts like 3D Object Detection, Expanded Kalman Filters, and 3D Fusion.
Or this one called Neural Optimization, teaching engineers advanced Deep Learning techniques to make their algorithms fast, optimized, and ready for deployment.
Or this course on Stereo Vision, showing engineers how to use 3D Geometries to reconstruct 3D Point Clouds, build Depth Maps, and more...
Our courses are unique because they're advanced.
When looking at the market around, you'll see many introduction-level products. Some courses will teach the fundamentals of Deep Learning, others will teach how to prompt to Chat-GPT. But no course really teaches how to build and train a HydraNet like Tesla has done, or how to switch a company's architecture to Transformer Networks.
Building advanced matters, because many engineers already have the fundamental skills, but still need to keep learning and advancing on their journey. If they stop learning, or if the learning path becomes too hard, then the company's knowledge asset stops growing.
ADVANCED LEVEL ONLY
BUILT WITH RESEARCHERS
EASY TO UNDERSTAND
Each course comes with a specific set of prerequisites — yet, these prerequisites often involve only a good understanding of Deep Learning Foundations, some prior experience with Computer Vision, and some Python skills. If your engineers can run an object detector, they already qualify to the programs.
The courses are already recorded and can be followed at your engineer's own pace. The format is a mix of text, videos, notebooks, code, and interviews with field experts.
There is no live version integrated, but you can make a request when purchasing.
One day, my team and I had a technical challenge to solve, and we didn't have the right skills to do it.
To be specific, it was a lane line detection challenge, and we were short on time.
So we started to think about outsourcing.
But to who?
It turns out, our CEO had a connection that could sell us the lane line detection implementation.
But there was a but.
After weeks spent with them, the algorithm was really bad.
It didn't detect lanes, and even less in our specific case.
Later on, we learned that it was because the outsourced company had been using an algorithm made from traditional OpenCV algorithms from the 90s, and that not only they didn't work, but they were terribly slow.
We had no real escape.
We were outnumbered, and most of our team didn't have any skills in this particular task.
But there's worse.
None of the companies out there could provide an algorithm based on Deep Learning (which was new and powerful at the time), and therefore...
We were stuck with the poorly performing algorithm!
Simply because the technology was new, and because we weren't trained enough to face the challenge.
The algorithm we had to purchase was based on traditional Computer Vision algorithm, and had no shot at winning against the emerging Deep Learning architectures.
We see events like this every day.
No action — Waiting for too long.
We hope the engineer will solve the issue — without a plan B. While in reality, having a trained team should be the only plan.