Computer Vision
CNNs, object detection, segmentation, and image augmentation techniques
Foundations
- Computer Vision
Image Classification
Given a fixed-size image, predict which class it belongs to from a predefined set, the task that launched the deep learning revolution when AlexNet cut the ImageNet error rate nearly in half in 2012.
- Computer Vision
Stereo Vision
Two cameras separated by a known baseline see the same point shifted by a disparity, and that shift triangulates depth with no active sensor.
- Computer Vision
Camera Calibration and Parameters
Calibration learns how one camera turns 3D world points into 2D pixels: where it sits, how it aims, and how its lens bends the light.
- Computer Vision
Color Spaces in Vision
Why raw RGB pixels are brittle under changing light, and how spaces like HSV or YCbCr separate color from brightness for robust vision.
- Computer Vision
Epipolar Lines Shrink Stereo Search
Two cameras looking at the same point are tied by geometry, so the match for any left pixel must sit on one specific line in the right image. Stereo search walks that line instead of the whole picture.
- Computer Vision
Frequency Domain Image Processing
Instead of viewing an image as a grid of pixels, frequency domain processing treats it as a sum of 2D sine and cosine waves, allowing us to easily filter out noise or isolate specific patterns.
- Computer Vision
Image Formation and Camera Model
How a 3D physical world is mathematically projected into a 2D pixel array, forming the foundation of computer vision and multi-view geometry.
- Computer Vision
Image Transformations and Homography
Affine maps resize, rotate, and shear images while keeping parallel lines parallel, and a homography adds perspective so tilted planes flatten back out.
- Computer Vision
Morphological Image Operations
Morphological operations are simple, non-linear techniques that process images based on shapes, allowing you to remove noise, separate touching objects, or find edges using a small sliding template called a structuring element.
- Computer Vision
Structure from Motion
Overlapping photos of one scene reveal both the 3D point positions and the camera path, because each matched feature triangulates from several viewpoints at once.
- Computer Vision
Visual SLAM
A camera dropped in an unknown building must draw the map and track itself inside it at the same time. Visual SLAM does both from images alone, then snaps the map shut when it recognizes a place it has seen.
- Computer Vision
Image Fundamentals
A digital image is a grid of numbers where each cell records how much light landed there. Resolution, channels and bit depth decide how much of the scene those numbers can hold.
- Computer Vision
Pixels
A pixel is one cell of the image grid holding a single light measurement. Millions of them tile into rows and columns, and their spacing sets the finest detail any image can carry.
- Computer Vision
RGB
RGB builds every screen color from three lights. Red, green and blue. Each pixel stores one brightness per light, and the triple sets the exact shade you see.
- Computer Vision
Grayscale Turns Color Into Brightness
Grayscale keeps one brightness number per pixel and drops the color. The standard mix weights green most, because human eyes respond to green light best.
- Computer Vision
Image Histograms
A histogram counts how many pixels sit at each brightness level. Piles against the left edge mean underexposure, spikes at the edges mean detail clipped away forever.
- Computer Vision
Image Resizing
Resizing remaps an image grid to a new width and height with scale factors, averaging neighborhoods to shrink and sampling between pixels to enlarge.
- Computer Vision
Image Interpolation for Resizing
Image interpolation estimates off-grid pixels when resizing or warping, choosing nearest, linear, cubic, area or Lanczos sampling by direction and budget.
- Computer Vision
Image Enhancement
Image enhancement remaps pixel values to raise contrast and visibility without adding information, grouping point operations, histograms and filtering.
- Computer Vision
Histogram Equalization Spreads Flat Images
Histogram equalization spreads clustered gray levels with the cumulative distribution so the output histogram sits flat and hidden detail shows.
- Computer Vision
Morphological Erosion Explained
Erosion keeps a foreground pixel only when a small probe fits fully inside the object. Shapes shrink, specks vanish, and thin bridges snap.
- Computer Vision
Morphological Dilation Explained
Dilation lights a pixel when any part of a small probe touches foreground. Shapes swell, holes plug, and near objects join hands.
- Computer Vision
Morphological Opening Explained
Opening erodes first to delete specks, then dilates to regrow the survivors. Noise leaves while true objects return to full size.
- Computer Vision
Morphological Closing Explained
Closing dilates first to plug holes, then erodes to pull outlines back. Gaps fill while outer boundaries return to place.
- Computer Vision
Fourier Transform for Images
The Fourier transform rewrites a photo as a recipe of waves. Slow shading becomes central low notes and crisp detail becomes outer high notes.
- Computer Vision
Camera Calibration with a Checkerboard
Calibration shows the camera a checkerboard of known size from many angles and solves for the numbers that explain every corner. The verdict is one score: mean reprojection error in pixels.
- Computer Vision
Bundle Adjustment for Maps and Poses
Initial cameras and points are roughly right but disagree by pixels. Bundle adjustment nudges all of them together until every 3D point lands on its photos.
- Computer Vision
Depth Maps from Single Images
A depth map stores one distance per pixel, which turns a photo into a 2.5D point cloud but leaves occlusions unknown.
- Computer Vision
DINOv2 Teacher Student at Scale
Train a student to copy a slow teacher on big and small crops, at both image and patch level. Scale plus curation makes frozen features excel.
- Computer Vision
Essential Matrix for Calibrated Views
The essential matrix is the calibrated cousin of the fundamental matrix: with intrinsics known, five point pairs recover the rotation and the direction of travel between two views, up to scale.
- Computer Vision
Fundamental Matrix of Two Camera Views
The fundamental matrix is the 3 by 3 table that encodes how two uncalibrated cameras sit relative to each other. It turns a pixel in one image into a search line in the other, and the true match always scores zero.
- Computer Vision
Lens Distortion: Radial and Tangential
Real lenses bend straight lines, so every photo carries a small warp. Lens distortion names that warp with a few coefficients, and undistortion removes it before you measure anything.
- Computer Vision
Masked Autoencoder Vision
Hide three quarters of an image and force the net to paint the missing pixels. Guessing so much from so little teaches shape and context.
- Computer Vision
MoCo Momentum Contrast
Keep a slow copy of the encoder and a long queue of old views. Each new view must pick its mate out of thousands of stored strangers.
- Computer Vision
Multi-View Geometry from Many Views
Two views give lines to search; three or more views pin a point down. Multi-view geometry is the shared rulebook for turning many photos into one 3D shape.
- Computer Vision
Multi-View Stereo for Dense Depth
One pair gives a noisy shell; many pairs fill the holes. Multi-view stereo fuses depth from several viewpoints into one dense cloud.
- Computer Vision
Novel View Synthesis from Photos
Novel view synthesis renders unseen viewpoints by interpolating learned scene structure instead of copying pixels.
- Computer Vision
Point Clouds as 3D Point Sets
Point clouds keep raw 3D coordinates with no grid and no surface, which saves memory but leaves neighbor search to you.
- Computer Vision
SimCLR Contrastive Pretraining
Show the net two random crops of one photo and teach it they match, while crops from other photos do not. Matching without labels builds strong features.
- Computer Vision
Sensor Fusion
Sensor fusion combines camera detail with lidar depth and radar velocity, so the car keeps working when any single sensor degrades.
- Computer Vision
Triangulation from Two Image Rays
Each camera turns a pixel into a ray in space. Triangulation is where those two rays come closest, and that closest point is the 3D answer.
- Computer Vision
Visual Odometry from Frame Motion
Watch the ground slide past frame by frame and add up the steps. Visual odometry turns pixel motion between consecutive frames into a running camera path.
- Computer Vision
Voxels for 3D Shape Grids
Voxels put 3D shapes on a fixed grid so 3D convolutions can run, at the cost of memory that grows cubically.
Detection And Localization
- Computer Vision
Object Detection
Object detection finds every instance of every object class in an image, returning a bounding box and class label for each — combining classification and localization in a single forward pass.
- Computer Vision
Anchor Boxes
Preset box shapes tiled across the image give the detector a starting guess, so it only predicts small nudges instead of raw coordinates from scratch.
- Computer Vision
Non Maximum Suppression
This concept covers the fundamentals of non maximum suppression within the broader context of Computer Vision.
- Computer Vision
Region Proposal Networks
A Region Proposal Network scans the shared feature map with ready-made anchor boxes and flags the few regions worth a closer look, so the detector never slides over empty background.
- Computer Vision
YOLO Architecture
YOLO frames object detection as a single regression problem, letting one network look at an image once and predict boxes and classes together.
- Computer Vision
Pyramids Detect Every Scale Once
A feature pyramid reuses the backbone once and fuses strong low-resolution semantics into high-resolution maps, so one detector reads tiny and huge objects together.
- Computer Vision
Classical Object Tracking
Classical trackers predict where a box moves next from its past motion, then correct with the new frame. That predict-then-correct loop holds one identity per object without running a detector on every frame.
- Computer Vision
Corner and Keypoint Detection
Corners and keypoints are distinct image regions that can be tracked across frames or matched between views, acting as stable anchors for vision tasks.
- Computer Vision
Face Detection and Recognition
Face recognition runs as a pipeline: a detector boxes each face, alignment standardizes the crop, and an embedding network names it by nearest neighbour.
- Computer Vision
HOG Features and Classical Detection
HOG describes shape by counting edge directions in small cells, so a pedestrian reads the same in sun or shadow without any learned weights.
- Computer Vision
Transformer-based Detectors
DETR predicts a fixed set of boxes and matches each truth exactly once, so anchors and NMS disappear; deformable attention makes it train fast by sampling a few points per scale.
- Computer Vision
Harris Corner Detector Finds Stable Points
Harris turns every pixel into a corner score from local gradients, so tracking keeps points that stay put when the camera shifts.
- Computer Vision
SIFT Scale Invariant Features
SIFT finds distinctive points that survive zoom and rotation, then fingerprints each one with a 128-number vector, so two photos of the same place still match from different viewpoints.
- Computer Vision
Oriented FAST and Rotated BRIEF
ORB fixes FAST with a pyramid plus orientation and pairs it with a rotated binary test, giving free and fast points for matching.
- Computer Vision
Feature Matching Links Two Views
Feature matching pairs each descriptor in one image with its nearest neighbor in the other, then keeps only the pairs that survive ratio and geometry checks.
- Computer Vision
Lowe's Ratio Test
The ratio test keeps a match only when the best candidate beats the runner up clearly, which drops pairs from repetitive texture.
- Computer Vision
RANSAC Robust Estimation
RANSAC fits models to noisy matches by voting: sample few points, count supporters, and keep the model with the largest consensus.
- Computer Vision
Homography Estimation Fits Planes
Homography estimation solves the 3 by 3 matrix that maps one plane to another, so tilted documents flatten and photo pairs align.
- Computer Vision
Contours and Shape Analysis
Once objects become outlines, every question about shape, size, corners, and identity reduces to math on ordered boundary points. This page maps the family.
- Computer Vision
Contour Detection
Join scattered edge pixels into ordered boundary loops, and decide up front whether holes and nested shapes appear or stay hidden.
- Computer Vision
Convex Hull
Shrink-wrap a contour in the smallest convex polygon that holds it. The gaps between wrap and shape reveal fingers, notches, and every concave secret.
- Computer Vision
Hough Transform
Stop hunting lines in the image and let every edge pixel vote for the lines that could pass through it. The busiest ballot boxes are the real lines.
- Computer Vision
Template Matching
Template matching finds an object by sliding its picture across the image like a stencil and keeping the spot where the pixels line up best.
- Computer Vision
Lucas-Kanade Optical Flow
Lucas-Kanade tracks a point by assuming its whole neighborhood moves together, turning many shaky single-pixel guesses into one steady motion vector.
- Computer Vision
Background Subtraction
Background subtraction keeps a running portrait of the empty scene and flags any pixel that stops matching it, turning a fixed camera into a motion sensor.
- Computer Vision
Kalman Filter Predicts Then Corrects
The Kalman filter tracks by predicting where the object should be from its motion, then splitting the difference between that prediction and the new measurement by trust.
- Computer Vision
Faster R-CNN Learns Its Own Proposals
Faster R-CNN replaces slow selective search with a Region Proposal Network that learns proposals on shared GPU features, unifying the whole detector into one network.
- Computer Vision
SSD Object Detector
SSD predicts boxes from several feature-map scales in one pass, so shallow layers catch small objects while deep layers catch large ones.
- Computer Vision
RetinaNet Detector
RetinaNet pairs a feature pyramid with focal loss, which down-weights easy background boxes so the rare foreground objects steer training.
- Computer Vision
Anchor-Free Detection
Anchor-free detection predicts boxes from points or keypoints instead of preset templates, removing thousands of hand-tuned anchors per image.
- Computer Vision
Intersection over Union Scores Box Overlap
Intersection over Union divides the overlap of two boxes by their combined area, giving one number between 0 and 1 for how well boxes agree.
- Computer Vision
Face Landmarks Pin 68 Points for Alignment
Face landmark models pin a fixed set of points to the eyes, brows, nose and jaw, turning a loose face box into measured geometry that later steps can warp, compare and decorate in one shared coordinate frame.
- Computer Vision
Object Tracking Across Frames
Object tracking keeps one steady identity on each moving target across video frames, connecting lonely per-frame detections into life stories.
- Computer Vision
Multi-Object Tracking Explained
Multi-object tracking detects everything in each frame and then solves the seating puzzle of which box continues which identity.
- Computer Vision
DeepSORT Appearance Tracking
DeepSORT remembers what each person looks like, so tracks survive the short occlusions where pure motion matching hands out new identities.
- Computer Vision
ByteTrack Low-Score Matching
ByteTrack keeps the low-confidence boxes other trackers trash, matching strong ones first and rescuing occluded targets with the leftovers.
- Computer Vision
OCR From Pixels to Text
OCR reads text out of photos in two moves: find where the words sit, then decode what the letters say, surviving fonts, blur and bent pages along the way.
- Computer Vision
Text Detection in Images
Text detection draws boxes around words in photos, coping with rotation, curvature and dense lines that ordinary object boxes cannot fit.
- Computer Vision
3D Object Detection
3D detection draws boxes with length, width, height and heading, not flat rectangles, lifting 2D finds into metric cuboids a planner can avoid.
- Computer Vision
6D Pose Estimation
6D pose estimation reports where an object sits and how it is rotated, as 3D position plus 3D orientation for grasping and assembly.
- Computer Vision
BEVFormer Camera 3D Detection
BEVFormer turns six camera views into one bird's-eye map by letting each map cell ask the right image pixels for evidence.
- Computer Vision
DETR Detects Sets Without Anchors
DETR treats detection as dealing a fixed hand of 100 guesses and pairing each truth with exactly one of them, so hand-tuned anchors and cleanup by NMS disappear.
- Computer Vision
Grounding DINO Detects What Words Name
Grounding DINO guides detector queries with text, so each query hunts a phrase-named object. Language picks which image spots become boxes.
- Computer Vision
Open Vocabulary Detection
Open vocabulary detection finds objects named in plain words instead of a fixed label list, scoring each image region against each phrase so new names work with no retraining.
Segmentation
- Computer Vision
Semantic Segmentation
Semantic segmentation gives every pixel a class label, producing a dense map where all pixels of one class share a label, down to exact object boundaries.
- Computer Vision
Instance Segmentation Splits Same-Class Masks
Instance segmentation gives every object its own pixel mask, so two overlapping people stay two instances instead of melting into one blob.
- Computer Vision
Panoptic Segmentation
Panoptic segmentation gives every pixel one class label and one instance ID, parsing countable things and uncountable stuff in a single non-overlapping map.
- Computer Vision
Pooling and Downsampling
Pooling summarizes each small patch of a feature map into one value, trading exact positions for cheaper maps and sturdier features.
- Computer Vision
Pose Estimation
Pose estimation finds body joints as image points and links them into a skeleton, turning pixels into a structure that reveals action.
- Computer Vision
Optical Flow
Optical flow estimates where each pixel moved between two video frames, turning a pair of still images into a field of motion arrows.
- Computer Vision
Video Understanding
A single frame cannot tell sitting down from standing up mid-motion. Video understanding reads how frames change over time, fusing what each frame shows with how pixels move.
- Computer Vision
Classical Image Segmentation
Classical segmentation splits foreground from background with pixel cutoffs, edge maps, and grown regions, no neural network involved. It works wherever labelled training data is scarce and every decision must stay explainable.
- Computer Vision
Encoder-Decoder Segmentation Architectures
A downsampling encoder reads what is in the image while an upsampling decoder recovers where it is, and skip bridges return the edges pooling lost.
- Computer Vision
Medical Image Analysis
Medical imaging requires extremely precise pixel-level analysis to locate tumors or organs, often using specialized architectures like U-Net that preserve fine spatial details.
- Computer Vision
U-Net and V-Net
U-Net passes fine spatial details directly from the encoder to the decoder via skip connections, enabling crisp pixel segmentation even with scarce data.
- Computer Vision
Image Thresholding
Thresholding turns a grayscale image into a binary mask by comparing each pixel to a cutoff, the simplest segmentation step before counting and contours.
- Computer Vision
Adaptive Thresholding
Adaptive thresholding computes a cutoff per pixel from its local neighborhood mean or Gaussian sum minus a constant, handling uneven light and shadows.
- Computer Vision
Otsu Thresholding
Otsu thresholding picks the cutoff that minimizes within-class variance across the histogram, the best automatic split for bimodal images.
- Computer Vision
Watershed Segmentation
Read brightness as hills and valleys, flood the valleys with water, and draw borders where floods from different valleys meet. Touching objects finally split apart.
- Computer Vision
Graph Cut Finds the Cheapest Split
Turn the image into a network where cutting cheap links separates foreground from background at the lowest total cost. The minimum cut is the segmentation.
- Computer Vision
GrabCut Cuts Objects Out From One Box
Draw one loose box around the object and GrabCut learns what foreground and background look like, then refines the border itself by repeating graph cut with color models.
- Computer Vision
Fully Convolutional Networks
Fully convolutional networks turned a classifier into a dense pixel labeller by swapping fully connected layers for 1x1 convolutions plus learned upsampling.
- Computer Vision
DeepLab v1 to v3 Plus
DeepLab keeps resolution high with spaced-out atrous filters, sees many scales at once with ASPP, and sharpens edges with a small decoder in v3 Plus.
- Computer Vision
Dice Loss for Segmentation
Dice loss scores the overlap between prediction and truth instead of counting pixels, so a tiny tumour still shouts loudly against a huge background.
- Computer Vision
Mask2Former Universal Masks
Mask2Former sharpens MaskFormer with masked attention, multi-scale decoding, and stronger matching, topping semantic, instance, and panoptic boards at once.
- Computer Vision
OpenPose Real-Time Skeletons
OpenPose finds every visible joint plus a field of limb-direction arrows in one network pass, then links joints into skeletons by checking which connections follow the arrows.
- Computer Vision
3D Human Pose Lifting
3D human pose lifts flat 2D skeletons into depth-aware joint positions, resolving the missing camera-axis coordinate with motion or body-shape priors.
- Computer Vision
Mask R-CNN
Mask R-CNN takes a two-stage detector and gives every detected box its own pixel silhouette, so overlapping objects of the same class stay separate.
- Computer Vision
nnU-Net Self-Configuring Segmentation
nnU-Net reads your dataset spacing and size, then configures its own U-Net depth, patches, and batches without hand tuning.
- Computer Vision
SegFormer Segments Without Positions
SegFormer pairs a hierarchical transformer encoder with a tiny MLP decoder, so dense pixel masks come from four feature scales with no positional encodings at all.
- Computer Vision
SAM Cuts Masks From Any Prompt
SAM encodes each image once, then turns points, boxes, or text into masks on demand, so one model segments new objects with no retraining.
Training Techniques
- Computer Vision
Image Augmentation
Image augmentation artificially expands the training set by applying random label-preserving transformations, teaching the model that these variations do not change the label and reducing overfitting.
- Computer Vision
Transfer Learning in Computer Vision
Transfer learning starts vision models from weights already trained on millions of images, then adapts them to a small task instead of learning edges from zero.
- Computer Vision
Feature Extraction Turns Pixels Into Clues
Feature extraction compresses millions of raw pixels into edges, textures and compact vectors, so a classifier learns from the clues instead of drowning in the grid.
- Computer Vision
Domain Adaptation for Vision
A model trained on sunny streets stumbles in the rain. Domain adaptation reuses its labeled sunny images plus unlabeled rainy ones to close that gap.
- Computer Vision
Few Shot Learns From Tiny Support
Few-shot classifies a query by its nearest class prototype from a handful of support images, while zero-shot replaces the support set with text descriptions in a shared space.
- Computer Vision
Dataset Preparation for Vision
Dataset preparation turns a folder of raw photos into clean, split, sized and normalized tensors, so training sees consistent inputs and the test score means something.
- Computer Vision
Image Dataloaders
A dataloader turns stored files into steady GPU-ready batches, hiding disk reads, decoding and padding behind parallel workers so the GPU never waits.
- Computer Vision
Fine Tuning Adapts Vision Gently
Fine-tuning first trains a new head on frozen features, then unfreezes the top backbone block with a tiny learning rate so high-level features bend without erasing.
- Computer Vision
Pretrained Vision Models
A pretrained vision model ships with backbone weights learned from millions of photos, so your classifier starts from real edges and textures instead of random noise.
- Computer Vision
ArcFace Angular Margin Loss
ArcFace trains face embeddings on the surface of a sphere and demands an angular safety gap between identities, which sharpens open-set matching.
- Computer Vision
Metric Learning for Embeddings
Metric learning teaches networks to lay out a space where similarity means closeness, so new classes slot in without retraining any classifier.
- Computer Vision
Contrastive Loss Pairs Training
Contrastive loss trains on pairs, pulling same-class embeddings together while pushing different-class pairs beyond a safety margin.
- Computer Vision
Triplet Loss Anchor Training
Triplet loss trains on triples of anchor, match and stranger, demanding the match beat the stranger by a margin every single time.
- Computer Vision
Siamese Networks
A siamese network runs two inputs through twin networks with shared weights, learning similarity itself instead of memorizing categories.
- Computer Vision
Hard Negative Mining Picks the Confusing Cases
Hard-negative mining spends each batch on the near-misses the model still gets wrong instead of the easy cases already sitting at zero loss.
- Computer Vision
CutMix Augmentation
Paste a square from photo B into photo A and mix the labels by area. The model sees real pixels from both scenes with an honest mixed tag.
- Computer Vision
Mixup Augmentation
Blend two photos and blend their labels with the same weight. The model learns smooth transitions instead of memorizing sharp borders.
- Computer Vision
RandAugment Augmentation
Skip the search. Pick N random operations at strength M for every image. Two knobs replace a thousand GPU hour policy hunt.
- Computer Vision
Self-Training Uses Confident Labels
Self-training teaches a new model from the old model's most confident predictions, growing the labeled set without growing the labeling bill.
- Computer Vision
Weakly Supervised Learning
Weakly supervised learning draws boxes and masks from image-level tags using heatmaps and multi-instance rules instead of hand-drawn outlines.
Convolutional
- Computer Vision
Convolutional Layers
Small shared filters scan the image and fire wherever their pattern appears, turning millions of pixels into stacked feature maps with few weights.
- Computer Vision
Spatial Filtering and Edge Detection
Sliding a small kernel across an image rewrites each pixel from its neighbors, so one pass can blur noise, sharpen detail, or expose edges as high gradient responses.
- Computer Vision
Basic CNN Operations
Convolutions slide small weight templates across an image to spot local patterns, while strides, padding, and pooling govern how feature maps shrink.
- Computer Vision
Early CNN Architectures
From handwritten digits in LeNet-5 to GPU-scaled AlexNet and uniform 3x3 stacks in VGG, each milestone proved deeper layers extract richer features.
- Computer Vision
Residual and Dense Networks
Instead of learning a full mapping from scratch, skip connections let layers learn only the residual difference, enabling networks hundreds of layers deep.
- Computer Vision
Gaussian Filter Smooths With a Bell Curve
A Gaussian filter blends each pixel with its neighbours, weighting close neighbours most by a bell curve, so noise fades while shapes stay fair.
- Computer Vision
Median Filter for Salt Pepper Noise
A median filter swaps each pixel for the middle value of its window. Lone specks get outvoted while true edges stand untouched.
- Computer Vision
Bilateral Filter Edge Smoothing
A bilateral filter blends only neighbours that are close and look similar. Flat zones smooth out while true edges refuse to mix.
- Computer Vision
Sobel Operator Edge Detection
Sobel pairs two small kernels that sense brightness change sideways and vertically. Strong responses mark edges, with direction thrown in.
- Computer Vision
Canny Edge Detector Stages
Canny chains smoothing, gradients, thinning, and smart thresholding into thin clean edges. Weak pixels survive only beside strong ones.
- Computer Vision
1x1 Convolution
A 1x1 convolution mixes channels at each pixel without touching neighbours, so it squeezes wide layers down, expands them back up, and adds depth cheaply.
- Computer Vision
Dilated Convolution Widens View Without Weights
Dilation spreads a small kernel over a wider area by skipping cells, so one layer sees far without extra weights and without shrinking the image.
- Computer Vision
Transposed Convolution
Transposed convolution learns how to upsample. Each input pixel stamps a small kernel onto a larger canvas, so decoders rebuild resolution step by step.
- Computer Vision
MobileNet Efficient Convolutions
MobileNet splits each convolution into a per-channel spatial pass and a 1x1 mix, cutting phone inference cost nearly ninefold while keeping accuracy close to full networks.
- Computer Vision
EfficientNet Compound Scaling
EfficientNet scales depth, width and resolution together with one knob, so each bigger model spends its extra budget where it helps instead of on depth alone.
- Computer Vision
ConvNeXt Modernized Convolutions
ConvNeXt rebuilds ResNet with transformer-era training and large depthwise kernels, proving tuned convolutions still match vision transformers on accuracy and speed.
- Computer Vision
Feature Maps Are Where Filters Fire
Each filter draws one map of where its pattern fired. Stack the maps and you hold the image as the network sees it, channels first.
- Computer Vision
Padding and Stride
Padding adds a border so the filter can center on edge pixels. Stride sets how far it jumps. Together they decide the output size.
Generative
- Computer Vision
Image Generation
Image generation is the process where models learn the underlying distribution of visual data to synthesize entirely new images from scratch, often guided by prompts.
- Computer Vision
Stable Diffusion
Stable Diffusion generates detailed images on a single GPU by denoising inside a tiny compressed latent space steered by frozen text embeddings.
- Computer Vision
Classifier Guidance Steers Diffusion
Classifier guidance steers a diffusion sampler with gradients from a noisy image classifier, pushing each step toward the requested class.
- Computer Vision
CycleGAN
CycleGAN translates between image domains with no paired examples by forcing round trips, horse to zebra and back, to land where they started.
- Computer Vision
Pix2Pix
Pix2Pix learns translation from perfectly paired examples, turning sketches into photos by combining adversarial sharpness with pixel-wise faithfulness.
- Computer Vision
StyleGAN
StyleGAN separates what to draw from how to render it, so coarse layers pick the pose while fine layers pick freckles, each steerable on its own.
- Computer Vision
Image Denoising
Image denoising separates true signal from random sensor noise, recovering clean photos by learning what noise looks like at every brightness level.
- Computer Vision
Image Editing
Instruction-driven editing rewrites photos from plain commands like make it sunset, changing pixels through language instead of masks and brushes.
- Computer Vision
Style Transfer
Feed-forward style transfer repaints any photo in a learned style in one pass, trading the slow per-image optimization of early methods for speed.
Evaluation
- Computer Vision
Adversarial Examples in Vision
A nudge to a photo so small you cannot see it can still flip what a model reports, because the nudge points exactly along the model's own gradients.
- Computer Vision
Top-K Accuracy
Top-k accuracy counts a prediction right when the true label lands anywhere in the model's k highest scores, forgiving confusion between near-duplicate classes.
- Computer Vision
Mean Average Precision
Mean Average Precision averages each class's precision-recall area across IoU thresholds, condensing a whole detector into the one number leaderboards quote.
- Computer Vision
FID IS KID Score Generated Images
FID and KID price how far generated features sit from real ones, while IS prices confidence plus variety on generated images alone without any real set.
- Computer Vision
IoU and Dice Score the Same Mask Twice
IoU and Dice read the same pixel overlap through two linked ratios, with Dice always scoring higher, so reports must convert before comparing.
- Computer Vision
PSNR SSIM LPIPS Image Quality
Three fidelity lenses for restored images: pixel error, structural match, and learned perceptual distance.
- Computer Vision
Uncertainty Estimation in Vision
A softmax score of 0.97 can still be a guess. Uncertainty estimation measures disagreement across predictions, so strange inputs route to humans before they cause harm.
Generative Vision Audio Multimodal
- Computer Vision
Image Captioning
Image captioning models act as a bridge between vision and language, taking an image as input and generating a coherent textual description of its contents.
- Computer Vision
Image-to-Image GANs
Instead of generating an image from random noise, image-to-image GANs translate an input image from one domain into another, like turning a sketch into a photo.
- Computer Vision
Super-Resolution and Image Restoration
Super-resolution rebuilds a sharp high-resolution image from a blurry or noisy one, learning what plausible detail looks like instead of just smoothing pixels.
- Computer Vision
Multimodal Learning Hub
Combine photos, words, sound, and video so answers use every signal. Joint models beat single senses on messy scenes.
- Computer Vision
LLaVA Visual Chat Tuning
Plug CLIP patches into a chat model through one matrix and tune on machine-made instructions. The bot then chats about photos.
- Computer Vision
SigLIP Sigmoid Pair Loss
Score every image-text pair on its own as match or not, instead of picking one winner per row. Small batches then train fine.
Representation Learning
- Computer Vision
Neural Radiance Fields
Instead of explicitly storing 3D shapes, NeRF trains a neural network to memorize a 3D scene and render new views from any angle.
- Computer Vision
Point Cloud Processing
A point cloud processor reads every 3D point on its own with one shared network, then pools the results into a single shape summary that never depends on point order.
Transformation
Classification
Training
Analysis
- Computer Vision
Face Verification Matches One Pair at a Time
Face verification answers one narrow question about a pair of photos, same person or not, by checking whether their embeddings land close enough together to clear a threshold.
- Computer Vision
Tracking Metrics MOTA IDF1 HOTA
Tracking metrics split honesty three ways, asking whether the tracker found everything, kept every identity, and drew tight boxes.
- Computer Vision
Text Recognition From Crops
Text recognition reads a cropped word strip left to right, turning pixel columns into characters without anyone cutting the letters apart first.
- Computer Vision
Action Recognition in Video
Action recognition names what someone is doing in a clip by reading appearance and motion together, since one frozen frame cannot tell sitting down from standing up.
- Computer Vision
Video Classification Basics
Video classification labels a whole clip with one tag, sampling frames smartly and pooling them so the majority verdict survives boring stretches.
- Computer Vision
Depth Estimation from Images Guide
Turn flat photos into meters per pixel. Depth estimation is the umbrella that splits into stereo, single-image, and filled-sparse methods.
- Computer Vision
Monocular Depth from One Image
One photo has no baseline, yet size, overlap, and haze hint at distance. Monocular depth learns those cues to guess meters from a single view.
- Computer Vision
PointNet and PointNet++ on Raw Points
Points arrive shuffled with no grid. PointNet reads each point alone then max-pools so order never matters; PointNet++ adds local neighborhoods.
- Computer Vision
RAFT Recurrent Flow Refinement
Look up motion, fix a little, repeat. RAFT keeps one high-resolution flow field and polishes it with a learned recurrent step.
- Computer Vision
Domain Generalization and Adaptation
Models trained in one visual domain fail in another. Adaptation uses target data to adjust, generalization prepares for unseen domains without any.
- Computer Vision
Vision Explainability Overview
Explanations answer which pixels the model used. Different methods answer at different resolutions, and the honest ones state their resolution upfront.
- Computer Vision
Zero-Shot Image Labels
Write each class as a sentence, embed them once, and label photos by nearest text. No training photos needed for new names.
- Computer Vision
Industrial Vision Fits the Line Takt
Industrial vision turns factory line speed into a vision specification: fix the camera, light, and trigger first, then fit the whole decide chain inside the takt budget.
Retrieval
- Computer Vision
Person Re-Identification Search
Re-identification finds the same person across cameras that never overlap, matching clothes and shape since faces are usually too small to help.
- Computer Vision
Image Retrieval Visual Search
Image retrieval answers queries with pictures, ranking a gallery by embedding distance so the closest match to your photo or sketch floats to the top.
Architecture
- Computer Vision
DeiT Trains Vision Transformers Fast
DeiT adds a distillation token that copies a CNN teacher, so vision transformers learn ImageNet in days instead of months.
- Computer Vision
Hybrids Stage Convs Before Attention
Hybrids run convolutions early for local bias and attention late for global range, beating either alone on data efficiency and compute cost.
Data Quality
- Computer Vision
COCO Dataset
Everyday scenes with boxes, masks and captions per image, the default ground for detection and segmentation work.
- Computer Vision
ImageNet Large Classification
Millions of varied web photos across a thousand classes, the pretraining source that made transfer learning the default.
- Computer Vision
MNIST Handwritten Digits
Tiny grayscale photos of handwritten digits with one label each, the standard first dataset for image classification before moving to harder data.
- Computer Vision
Data Labeling Tools
Labeling tools turn raw photos into trusted training data with drawing UX, double review, and agreement scores that decide what ships.