Computer Vision
Computer Vision on AI-ML Companion: Vision tasks, what each one costs to label, and how each is scored. 10 interactive modules with live visualizations, quizzes, and hands-on Python coding.
Start free: What Computer Vision Actually Solves is fully open to everyone, no account required. The other 9 modules are part of AI-ML Companion Premium; every title and summary is listed below so you can see exactly what the track covers before deciding.
Modules in this track
- What Computer Vision Actually Solves (free) - Five tasks, what each output shape costs to label, and how each is scored
- Images, Labels, and the Data You Actually Get (premium) - What survives the trip from camera to tensor, and what caps your accuracy
- Vision Transformers (premium) - The prior you give up, the data that replaces it, and why detection needed windows
- Choosing a Backbone (premium) - Why the recipe beats the architecture, and why FLOPs are not milliseconds
- Object Detection (premium) - Proposals, the heuristic nobody trained, and what one mAP number hides
- Segmentation (premium) - Per-pixel answers, the location the encoder threw away, and why the same error scores twice
- Contrastive and Self-Supervised Vision (premium) - Free supervision, a billion comparisons, and the space that replaced the classifier
- Generative Vision in Production (premium) - The step count is the bill, the strength dial is a step count, and FID cannot see your image
- Video Understanding (premium) - Nine hundred images per clip, a frame rate that is really a correctness dial, and the metric that misses
- Depth, Geometry, and 3D (premium) - Projection destroyed the depth, one image cannot know the scale, and resizing invalidates your camera