From pixels to objects, scenes, and actions: the representations and architectures that power visual intelligence.
A learnable kernel scans local neighborhoods to detect reusable visual patterns.
A reusable technical reading framework for moving from vocabulary to working understanding. Use it alongside the formulas and papers below.
From pixels to objects, scenes, and actions: the representations and architectures that power visual intelligence. Start by naming the representation, objective, and source of evidence in any system you study. This habit separates a useful model explanation from a list of buzzwords.
Use the learning outcomes as checks for understanding: Choose between classification, detection, segmentation, and generation; Understand convolutional and patch-based vision models; Design evaluation around geometry, calibration, and shift. For each claim, ask what assumptions make it true, what data it needs, and how it could fail.
Translate each equation into a small experiment before treating it as memorized knowledge. 2D convolution: A learnable kernel scans local neighborhoods to detect reusable visual patterns.
Read primary work with a repeatable lens: problem, method, data, measurement, limits, and what would change your mind. This section starts with “ImageNet Classification with Deep Convolutional Neural Networks”, “U-Net: Convolutional Networks for Biomedical Image Segmentation”, “An Image is Worth 16x16 Words”. Follow citations outward only after you can explain the central claim in your own words.
Implement a minimal example using 2D convolution; record inputs, outputs, and a failure case.
Define a baseline, a success metric, a stress test, and the evidence required before you trust an improvement.
Choose one paper from the library, reproduce a small claim, then write down the questions its evaluation leaves open.
Primary sources and peer-reviewed work to deepen this page.
A. Krizhevsky · I. Sutskever · G. Hinton
AlexNet demonstrated the impact of GPU-trained deep CNNs on large-scale visual recognition.
O. Ronneberger · P. Fischer · T. Brox
A symmetric encoder–decoder with skip connections that made precise biomedical segmentation practical.
A. Dosovitskiy et al.
Shows that a pure Transformer over image patches can compete with convolutional vision models.