LibreMobileNetV4s-cls

LibreMobileNetV4s-cls

MobileNetV4-conv-Small image classifier (224px input, ImageNet-1k, 1000 classes), repackaged for LibreYOLO. ~3.8M parameters.

Source

Derived from huggingface/pytorch-image-models (timm), model mobilenetv4_conv_small.e2400_r224_in1k. Architecture and pretrained weights by Ross Wightman and the timm contributors (ImageNet-1k, trained in timm; no distillation, no extra data). Copyright (c) Ross Wightman. Licensed under the Apache License 2.0.

MobileNetV4 paper: "MobileNetV4 - Universal Models for the Mobile Ecosystem" (https://arxiv.org/abs/2404.10518).

Modifications

Learned parameters are unchanged from timm. The checkpoint is metadata-wrapped into the LibreYOLO format (model_family / task / nc / names). LibreYOLO's native MobileNetV4 implementation mirrors timm's module names, so inference is bit-identical to timm (max_abs_diff == 0, verified by the parity test). See weights/convert_mobilenetv4_weights.py in the LibreYOLO source repository.

Usage

from libreyolo import LibreYOLO

model = LibreYOLO("LibreMobileNetV4s-cls.pt")
result = model.predict("image.jpg")[0]
print(result.probs.top1, result.probs.top5)

License

Apache License 2.0. See the LICENSE and NOTICE files in this repository.