Computer Vision
30 bites tagged Computer Vision — interview questions with model answers, and 60-second explainers.
What does N-way K-shot classification mean?
N is classes per episode, K is labeled examples per class in the support set, prediction is on a separate query set. few-shot evaluation vocabulary. confusing K with total training data or swapping N and K.
How is IoU computed and why prefer mIoU?
IoU is intersection over union of predicted and true pixels; mIoU averages per class; pixel accuracy is dominated by background. segmentation metrics under imbalance. equating accuracy with IoU.
Sparse vs dense optical flow and Lucas-Kanade
Sparse flow tracks selected feature points, dense flow computes a vector per pixel; Lucas-Kanade solves brightness constancy in a local window assuming constant motion. understanding motion estimation granularity.
U-Net architecture and its skip connections
U-Net has a contracting encoder, an expanding decoder, and skip connections that concatenate matching-resolution encoder features into the decoder to recover spatial detail lost in downsampling. encoder-decoder design for segmentation.
Semantic versus instance segmentation
Semantic segmentation labels each pixel by class but merges objects of the same class; instance segmentation also separates individual objects. distinguishing two pixel-labeling tasks.
Detector head losses: regression versus classification
The head splits into a classification branch using cross-entropy over classes and a regression branch using a robust Smooth L1 or IoU loss on box offsets, combined as a weighted sum. multi-task loss design in detection heads.
Deploying real-time detection on edge devices
Pick an efficient one-stage detector, train with augmentation, then quantize, prune, and compile to a hardware-accelerated runtime, measuring latency and accuracy tradeoffs. end-to-end edge deployment reasoning.
Focal Loss and class imbalance in detectors
Focal loss multiplies cross-entropy by a (1-p)^gamma factor that down-weights easy, well-classified examples so the vast easy background does not swamp the loss. handling extreme class imbalance.
Mean Average Precision in object detection
AP is the area under the precision-recall curve per class; mAP averages AP over classes, and COCO also averages over IoU thresholds. the headline detection metric. confusing mAP with plain accuracy or ignoring the PR curve.
Intersection over Union for detection
IoU is the area of overlap divided by the area of union of predicted and ground-truth boxes; a threshold decides true positives. the core overlap metric.
Image classification versus object detection
Classification assigns one label to the whole image; detection localizes and labels multiple objects with bounding boxes and class scores. basic task definitions.
Translation equivariance versus invariance in CNNs
Convolution is equivariant, shifting input shifts feature maps; invariance comes only from pooling and global aggregation. Strict invariance is partial and broken by strided sampling. precise reasoning about CNN symmetries.
Depthwise separable convolution cost savings
Separable conv splits standard conv into per-channel spatial filtering plus a 1x1 pointwise mix, cutting cost by roughly 1/N plus 1/k². efficient convolution factorization.
Adapting a classification CNN for segmentation
Replace the dense head with conv layers, upsample via transposed convolutions, and fuse encoder skip connections to recover spatial detail lost to downsampling. turning a classifier into a dense predictor.
Uses of the 1x1 convolution
A 1x1 conv is a per-pixel linear combination across channels; it reshapes channel depth cheaply and adds nonlinearity. Uses: dimensionality reduction in bottlenecks and channel mixing. channel-wise operations and efficient design.
Receptive fields in convolutional networks
Receptive field is the input region affecting a neuron; it grows with depth, larger kernels, and stride. It matters for capturing context in detection and segmentation. how spatial context accumulates in CNNs.
ResNet residual blocks and the degradation problem
A residual block learns F(x) and adds the identity input x, so layers fit a residual; this eases gradient flow and solves the degradation problem. why skip connections enable very deep nets.
Regularization techniques for an overfitting CNN
Data augmentation expands the effective dataset, dropout prevents co-adaptation, weight decay penalizes large weights, plus early stopping and batchnorm. practical remedies for overfitting and their mechanisms.
Why CNNs need nonlinear activations like ReLU
ReLU introduces nonlinearity letting stacked layers model complex functions; without it any stack collapses to a single linear map. why nonlinearity matters in deep networks.
Max pooling versus strided convolution
Pooling downsamples and adds small translation invariance with no parameters; strided conv learns its downsampling but adds parameters. downsampling tradeoffs in CNNs.
How a convolutional layer works
Learnable kernels slide over the input computing dot products, with stride controlling step size and padding controlling output size. the mechanics of convolution.
Stereo rectification math and its artifacts
Rectification warps both images by homographies so epipolar lines become horizontal and aligned. epipolar geometry and stereo correspondence.
The PnP problem in Structure from Motion
PnP recovers a camera's pose from known 3D points and their 2D projections; it registers new frames against the existing point cloud in SfM. geometric camera pose estimation. confusing it with triangulation.
Walk me through a CNN's layers for image classification
Tests hierarchical feature extraction in CNNs. Answer: conv filters learn edges-to-objects with shared weights, pooling reduces dimensions and adds invariance, fully-connected layers classify.
Get Computer Vision bites daily.
Five a day, five minutes, offline. With quizzes so it sticks.
Open testing — you’ll join as an early tester.