CSATv2

Category: Hybrid · Code: timm/models/csatv2.py · Architectures: 2 · Pretrained weights: 3 (0 with ImageNet-1k results)

CSATv2 is a frequency-domain vision model that works on DCT transforms of the input image, with spatial attention.

Papers

The module docstring does not reference any papers.

Usage

See the Quickstart for loading pretrained weights, image preprocessing and fine-tuning, and Feature Extraction for using the models as backbones.

>>> import timm
>>> model = timm.create_model('csatv2.r512_in1k', pretrained=True)
>>> data_config = timm.data.resolve_model_data_config(model)

Model API support

Feature	`CSATv2`
`features_only=True` feature maps	✓ (strides 8, 16, 32, 64)
`forward_intermediates()`	✓
`prune_intermediate_layers()`	✓
Any input size, no changes needed	✓
`set_input_size()`	–
`dynamic_img_size=True`	–
`set_grad_checkpointing()`	✓
Layer-wise LR decay (`group_matcher()`)	✗
`reset_classifier()`	✓

✓ supported, ✓ (square) works for square sizes only, ✗ not supported, – not needed (the model already accepts any input size). Checked by running each feature on a randomly initialized model (CSATv2 with csatv2 at 256), other architectures of the same class are expected to behave the same. See the Model API page for details.

Architectures

Architecture	Params (M)	GMACs	Input size	Pretrained weights
`csatv2`	11.10	1.39	512	1
`csatv2_21m`	20.70	4.72	640	2

GMACs and parameter counts are for the default input size and classifier of each architecture, from the benchmark results.

Pretrained weights

Top-1 is ImageNet-1k validation accuracy from results-imagenet.csv at the listed eval size, - if the weight has no ImageNet-1k result (e.g. pretraining or image-text weights). Classes is the size of the pretrained classifier, none for weights without one.

Weight	Input size	Top-1 (%)	Eval size	Classes	License	Hub
`csatv2.r512_in1k`	512	-	-	1000	-	timm
`csatv2_21m.sw_r512_in1k`	512	-	-	1000	-	timm
`csatv2_21m.sw_r640_in1k`	640	-	-	1000	-	timm

Update on GitHub