You can do spot checks on production data by labeling some small amounts and comparing the performance to the performance on the test dataset.
This is of course a very expensive and error prone approach (see Simpson's paradox).
2/6
You can evaluate some high-level statistics about the performance of your application. Usually the CV model is not the last part of the pipeline.
For example, a car recognizing speed limits may check if the driver actually drives at a speed close to the speed limit.
3/6
Another example - you are building a web tool that removes the background from images. You can measure how many corrections the user makes manually after that.
In general, check if your customer uses your model output directly or corrects it somehow.
4/6
Another way is to train an autoencoder on the original dataset and note the reconstruction loss. After that, pass a dataset collected from production and see if the reconstruction loss is larger. This would indicate concept drift.
In summary, try to get as much information as possible from your end application. In the end, only issues that affect your final product are worth looking into!
6/6
• • •
Missing some Tweet in this thread? You can try to
force a refresh
Here is a list of useful courses if you want to learn about software for self-driving cars.
Some of the courses are paid, but all platforms offer regular discounts and financial aids if you can't affor them.
Thread 👇
Udacity Self-Driving Car Nanodegree
This program offers hands on experience on all kind of relevant topics like perception, localization, planning and control. It takes a lot of time, but it is worth it.
I plan many threads on self-driving cars and how to get into the industry.
I will link all of the individual threads that will be focused on a particular topic below.
🧵
I recently gave a talk on AI for self-driving cars for a @DeepLearningAI_ Pie & AI event hosted by @Jeande_d. You can check out the recording on YouTube.
I will be posting threads summarizing the talk and link them below 👇
CNNs are an important class of deep artificial neural networks that are particularly well suited for images.
If you want to learn the important concepts of CNNs and understand why they work so well, this thread is for you!
🧵👇
What is a CNN? 🤔
A CNN is a deep neural network that contains at least one convolutional layer. A typical CNN has a structure like this:
▪️ Image as input
▪️ Several convolutional layers
▪️ Several interleaved pooling layers
▪️ One/more fully connected layers
Example: AlexNet
A good example - AlexNet
Throughout the thread I will be giving examples based on AlexNet - this is the net architecture that arguably started the whole deep learning revolution in computer vision!