Skip to content
tezvyn:

All bites

The whole library, newest first. Filter by what you are here for, or pick a topic if you already know.

4330 bites

Page 85

Computer Vision1 min read

Brightness constancy and small-motion assumptions

Brightness constancy says a point's intensity is invariant under motion; small motion lets you linearize via Taylor expansion.

Computer Vision1 min read

Kalman filter for bounding-box tracking

State, transition, measurement models, and process plus measurement noise; predict then correct each frame. State holds box position and velocity; measurement is the detected box.

Computer Vision1 min read

3D CNNs vs two-stream action recognition

3D CNNs learn spatiotemporal filters end to end but are heavy; two-stream splits RGB appearance and precomputed optical flow, strong but costly to compute flow.

Computer Vision1 min read

Re-identification in multi-object tracking

Re-ID matches an object to its prior id using appearance embeddings, not just position; store track features and match re-entering detections by embedding similarity.

Computer Vision2 min read

Self-supervised pretraining for video understanding

Define a label-free task like temporal order prediction or contrastive clip matching that forces temporal reasoning, then fine-tune on labeled action data.

Computer Vision2 min read

Scene flow versus optical flow

Optical flow is 2D pixel motion in the image plane; scene flow is the 3D motion field of points in space, needing depth via stereo, RGB-D, or LiDAR.

Computer Vision2 min read

How ViT and CNN process an image differently

A CNN slides local filters over the raw pixel grid; a ViT splits the image into patches, flattens and linearly embeds each into a token, adds positional embeddings, and feeds the sequence to…

Computer Vision1 min read

Self-attention over image patches explained

Each patch projects to query, key, value; a patch's query is scored against all keys, softmax-normalized into weights, used to combine all values.

Computer Vision1 min read

Inductive biases of ViT versus CNN

CNNs bake in locality and translation equivariance; a plain ViT has almost none beyond patch structure, so it must learn spatial relations from data, needing large datasets or strong pretraining.

Computer Vision2 min read

How Swin Transformer achieves linear attention

Swin computes attention within local non-overlapping windows of fixed size, making cost linear in patches, then shifts windows between layers so information crosses boundaries.

Computer Vision1 min read

Cross-attention for visual question answering

In cross-attention queries come from one modality and keys/values from the other, e.g. text queries attend over image features so the question selects relevant regions.

Computer Vision2 min read

Why ViTs need positional embeddings

Self-attention is permutation invariant so patch order is lost; positional embeddings restore spatial location. CNNs encode position implicitly via the fixed convolution grid.

Computer Vision2 min read

Core principles of a Neural Radiance Field

An MLP maps a 3D point plus view direction to color and density; novel views render by casting rays, sampling points, querying the MLP, and volume-integrating along each ray.

Computer Vision2 min read

Pure ViT vs hybrid CNN-Transformer for medical segmentation

Pure ViT captures global context but is data hungry and weak on local detail; hybrid CNN-Transformer gets local features cheaply plus global attention, ideal for scarce…

Computer Vision2 min read

Attention in diffusion U-Nets for text conditioning

Self-attention mixes spatial features at low-res blocks; cross-attention has image queries attend to text-token keys/values; placed inside transformer blocks.

Computer Vision1 min read

GAN architecture: generator and discriminator roles

Generator maps noise to fake samples, discriminator classifies real versus fake, they train as a two-player game until samples fool the discriminator.

Computer Vision2 min read

Diffusion forward and reverse processes

Forward process gradually adds Gaussian noise until data is pure noise; reverse process learns to denoise step by step; the network predicts the noise added at each timestep.

Computer Vision2 min read

Mode collapse in GAN training

Generator produces few outputs ignoring data diversity, caused by chasing whatever fools the current discriminator; mitigate with minibatch discrimination, unrolled GANs, or Wasserstein loss.

Computer Vision2 min read

Evaluating generative models with FID versus IS

FID compares Inception feature distributions of real and fake images via Frechet distance between two Gaussians; it uses real data as reference and detects diversity issues, unlike IS which uses no real…

Computer Vision1 min read

Why U-Net skip connections matter for denoising

Skips carry high-resolution spatial detail from encoder to decoder, preserving fine structure lost in downsampling and easing gradient flow, which lets the model restore detail while removing…