EIDORA/AlexNet_IN1k

AlexNet Image Light

AlexNet is a convolutional image model from the ImageNet classification era. This EIDORA adaptation extracts a 4096-dimensional feature vector from its classifier rather than returning class probabilities.

Training Data and Checkpoint Provenance

• Architecture: AlexNet convolutional network with fully connected classifier.
• Original checkpoint: torchvision/alexnet-imagenet1k-v1.
• Weight variant: IMAGENET1K_V1 (TorchVision download).
• Training data: ImageNet-1K
• Training method: Supervised image classification.
• Source: AlexNet (original author attribution retained).

Model Overview

• Model: AlexNet Image Light
• EIDORA identifier: eidora/image-alexnet-imagenet-light
• Modality: Image
• Embedding dimensions: 4096
• Embedding normalization: L2-normalized
• Similarity measure: Cosine similarity
• Package schema: Version 3
• Model version: 0.1.0

EIDORA Use and Supported Modalities

Supported modality: Images only.

Researchers can use this package to:

• Use a familiar CNN baseline when comparing image embedding models.
• Explore simple visual similarity and clustering in an image collection.
• Compare older CNN features with ResNet, DINO and CLIP.

Input and Preprocessing

EIDORA prepares the image before inference:

1. Convert to RGB.
2. Resize the shorter side to 256 pixels, then take a center crop of 224 × 224.
3. Use bilinear interpolation. Use truncate dimension rounding.
4. Rescale by 0.00392156862745098 and normalize each RGB channel.
5. Arrange the tensor in NCHW layout and pass float32 pixel_values with shape [batch, 3, 224, 224].

Mean: [0.485, 0.456, 0.406]
Standard deviation: [0.229, 0.224, 0.225]

Normalization is performed by EIDORA and should not be applied a second time. The defaults in config.yaml determine the documented embedding behavior; alternate settings may change the vectors.

Output Embeddings

The ONNX output is float32 with shape ['batch', 4096]. Each vector has 4096 dimensions, is L2-normalized, and is intended for cosine similarity comparisons.

EIDORA retains the pretrained image network and extracts features from the second fully connected layer after ReLU, followed by L2 normalization.

Runtime and Compatibility

The package uses model.onnx and schema-v3 runtime declarations for image preparation and embedding output.

It declares CPUExecutionProvider. Actual execution availability depends on the installed ONNX Runtime and EIDORA Desktop environment. config.yaml is authoritative for graph inputs, preprocessing, provider and embedding dimensions.

Limitations

• The model was trained for ImageNet classification, not cultural or historical interpretation.
• CNN feature proximity does not establish common authorship, identity or meaning.
• Other image resize settings may produce different embeddings.

Results should be evaluated against the intended research data and questions.

Research Context and References

Main paper

ImageNet Classification with Deep Convolutional Neural Networks

Paper: https://proceedings.neurips.cc/paper/2012/hash/c399862d3b9d6b76c8436e924a68c45b-Abstract.html

The publication describes the upstream learning method and original model. The embeddings returned by EIDORA additionally depend on the specific preprocessing and output adaptation documented here.

Primary citation (BibTeX)

@inproceedings{krizhevsky2012imagenet,
  title={ImageNet Classification with Deep Convolutional Neural Networks},
  author={Krizhevsky, Alex and Sutskever, Ilya and Hinton, Geoffrey E.},
  booktitle={Advances in Neural Information Processing Systems},
  year={2012}
}

License and Attribution

The TorchVision software library uses BSD-3-Clause terms. This does not by itself assign those terms to the pretrained ImageNet weights. The linked TorchVision model documentation does not establish a separate license for these pretrained weights. No weight-license identifier is asserted here. The ONNX graph contains converted weights from TorchVision. EIDORA does not claim authorship of the original model.

No separate upstream checkpoint LICENSE file was supplied in the captured EIDORA repository; that absence should not be interpreted as permission or prohibition.

Package Information

• Model repository: EIDORA/AlexNet_IN1k
• Package version: 0.1.0
• Package schema: 3
• EIDORA configuration: config.yaml
• ONNX graph(s): model.onnx
• Supported modality: Image
• License identifier: other
• Declared provider: CPUExecutionProvider

The configuration defines the authoritative model-facing runtime contract. Other packaged assets, where present, do not introduce additional inference modalities.