Project AURIONAI Use-Cases Research & Innovation
Google
vision

EfficientNetV2

Second EfficientNet generation with faster training, from the paper "EfficientNetV2: Smaller Models and Faster Training".

GoogleGoogleUSAEfficientNetFamily
Figure from the EfficientNetV2 paper: Training process in our improved progressive learning - It starts with small image size and weak regularization (epoch=1).
Levelvariant
Classvision
Parameters21.61M
Contextnot in the notes
Released1 Apr 2021
Opennessopen weights permissive
Runs onmobile or NPU
Statusactive

Lineage

successor
EfficientNet-B0 to EfficientNetV2: successor, high confidenceEfficientNet-B0EfficientNet-B0Google, May 2019EfficientNetV2EfficientNetV2Google, Apr 2021
GoogleEfficientNet-B0 is the parent (successor)high confidencenotes
Where this comes from
Parent and child relations from model cards, Hugging Face base_model fields and the genealogy tables of the research notes. The tree shows two generations each way, 2 models. Confidence and evidence are the notes' own.

R&D Topics

Efficient vision networks for mobile and edge devicesproduced by

Identity

Developer
Google
Family
EfficientNet
Released
1 Apr 2021 sourcesannouncement page
Weights published
13 Dec 2022
Record checked
12 Oct 2026
Where this comes from
Developer and dates from the research notes. Release date is the vendor's public announcement; weights date is when open weights were published.

Architecture and Size

Architecture
CNN
Total parameters
21.61M sources
Active parameters
–
Checkpoint parameters
–
Note
B0 to B3, S, M, L, XL.
Context length
not in the notes
Input
image
Output
labels and boxes
Where this comes from
Parameter counts as the developer states them. Where the checkpoint holds a different total, both are shown (register A16).

Openness and Access

Openness
open weights permissive
Licence
Apache License 2.0 sources
Access routes
open weights download, cloud marketplace
Where this comes from
The licence file governs over registry tags and aggregator labels.

Deployment

Runs on
mobile or NPU
Footprint
Edge or toolkit builds listed in: Qualcomm AI Hub.
Runtimes
not in the notes
Where this comes from
Deployment tiers and runtimes as named in the notes. Footprint is the notes' figure, not a measurement by AURION.

Indonesia

Indonesian support
not stated
Regional languages
not in the notes
Jakarta hosting
unknown
Languages
not in the notes
Where this comes from
Indonesia fields are carried on every model; unknown is an allowed value.

Benchmark Results

The notes record no benchmark result for this model.

Industry Use Cases

No industry use case links this model.

Sources

  1. arxiv.org/abs/2104.00298 https://arxiv.org/abs/2104.00298
  2. huggingface.co/timm/tf_efficientnetv2_s.in21k_ft_in1k https://huggingface.co/timm/tf_efficientnetv2_s.in21k_ft_in1k
  3. jumpstart-cache-prod-us-west-2.s3.us-west-2.amazonaws.com/models_manifest.json https://jumpstart-cache-prod-us-west-2.s3.us-west-2.amazonaws.com/models_manifest.json
  4. aihub.qualcomm.com/models/efficientnet_v2_s https://aihub.qualcomm.com/models/efficientnet_v2_s
  5. github.com/pytorch/vision/blob/main/torchvision/models/e... https://github.com/pytorch/vision/blob/main/torchvision/models/efficientnet.py
  6. pytorch.org/vision/stable/models/efficientnetv2.html https://pytorch.org/vision/stable/models/efficientnetv2.html
  7. github.com/huggingface/pytorch-image-models/blob/main/RE... https://github.com/huggingface/pytorch-image-models/blob/main/README.md

To cite this page: Project AURION, AI Use-Cases Research & Innovation, "EfficientNetV2", https://aurion.semai.id/models/google-efficientnetv2/, records of October 2026.