ImageNet 1K: 32x32 Downsampled Image Classification Dataset

ImageNet 1K 32X32 is a repackaged version of the ILSVRC/ImageNet-1k dataset. The images have been center-cropped to square and resampled to a 32x32 pixel resolution using Lanczos interpolation. The dataset was created by benjamin-paine and last updated on Hugging Face in September 2024.

Use Cases

• Benchmarking image classification models based on the 32x32 pixel image format.
• Training efficient neural networks for resource-constrained environments based on the downsampled image size.
• Studying the effects of image resolution on model performance based on the described resampling process.

Strengths

• Images are preprocessed to a consistent 32x32 pixel square format.
• Derived from the well-known ILSVRC/ImageNet-1k benchmark dataset.
• Last updated on the platform on 2024-09-15.

Limitations

• Row count, column details, and license information are unknown.
• The original ImageNet data may reflect geographic and cultural biases inherent in its source collection.
• The extreme downsampling to 32x32 pixels may limit suitability for tasks requiring fine-grained visual detail.

Provenance

Source

ILSVRC/ImageNet-1k

Collection Method

Images were center-cropped to square and resampled to 32x32 pixels using Lanczos interpolation.

Time Range

null

Freshness

Last updated 2024-09-15 03:18:22.

Geography

null