TinyImageNetDataLoaderOptions
Options & Configuration · AiDotNet.Data.Vision.Benchmarks
Configuration options for the Tiny ImageNet (200-class, 64×64) data loader.
How It Works
Tiny ImageNet is the standard middle-ground vision benchmark between CIFAR-100 and full ImageNet — 200 classes with 500 training, 50 validation, 50 test images each at 64×64 resolution. Used widely for architecture-search and few-shot studies. Produced by Stanford CS231n.
Constructors
| Constructor | Summary |
|---|---|
TinyImageNetDataLoaderOptions() |
Properties
| Property | Summary |
|---|---|
DatasetSplit Split { get; set; } |
Dataset split to load. |
bool AutoDownload { get; set; } |
Automatically download if not present. |
bool Normalize { get; set; } |
Normalize pixel values to [0, 1]. |
int ImageSize { get; set; } |
Image size (height = width). |
int? MaxSamples { get; set; } |
Optional maximum number of samples to load. |
string DataPath { get; set; } |
Root data path. |