CanViT: Toward Active-Vision Foundation Models (NeurIPS 2026)

Yohaï-Eliel Berreby^1,2 · Sabrina Du^1,2 · Audrey Durand^2,3 · B. Suresh Krishna^1
1McGill University · ²Mila – Quebec AI Institute · ³Université Laval

This repository holds the reference PyTorch implementation of CanViT, with pretraining, task specialization and evaluation; the package is canvit-pytorch on PyPI.

News

• 2026-09-26: canvit-pytorch 0.2, a refactoring release; updated checkpoints pushed to the Hub. Code written for 0.1: see Troubleshooting.
• 2026-09-24: 🎉 Accepted at NeurIPS 2026!
• 2026-05-16: Preprint v2 (arXiv:2603.22570v2), adding the 84.5% ImageNet-1k fine-tuning result and the effect of canvas resolution.
• 2026-04-06: First finetuned IN1k checkpoint: canvitb16-add-vpe-finetune-g128px-s512px-in1k-2026-04-06, with new CanViTForImageClassification API.
  • 🎉 CanViT sets a new SOTA on active-vision IN1k classification, with 84.5% top-1 accuracy, up from AdaptiveNN's previous best of 82.2%.
• 2026-03-23: Preprint v1 (arXiv:2603.22570v1).
  • 🎉 CanViT sets a new SOTA on active ADE20K segmentation, with 45.9% ADE20K mIoU, obtained using linear probing from frozen weights.
• 2026-02-18: Initial code and first pretrained checkpoint release.

Abstract

Active computer vision promises efficient, biologically plausible perception through sequential, localized glimpses, but lacks scalable general-purpose architectures and pretraining pipelines, leaving Active-Vision Foundation Models (AVFMs) underexplored. We introduce CanViT, the first task- and policy-agnostic AVFM. CanViT uses scene-relative RoPE to bind a retinotopic Vision Transformer backbone and a spatiotopic scene-wide latent workspace, the canvas. Efficient interaction with this high-capacity working memory is supported by Canvas Attention, a novel asymmetric cross-attention mechanism. We decouple thinking (backbone-level) and memory (canvas-level), eliminating canvas-side self-attention and fully-connected layers to achieve fast sequential inference and scalability to high output resolutions. We propose a label-free active vision pretraining scheme, policy-agnostic passive-to-active dense latent distillation: reconstructing scene-wide DINOv3 embeddings from sequences of low-resolution glimpses with randomized locations, zoom levels, and lengths. We pretrain CanViT-B from a random initialization on 13.2 million ImageNet-21k scenes—an order of magnitude more than previous active models—and 1 billion random glimpses, in 166 hours on a single H100. On ADE20K segmentation, a frozen CanViT-B achieves 38.5% mIoU in a single low-resolution glimpse, outperforming the best active model's 27.6% with 20x fewer inference FLOPs as well as its FLOP- or input-matched DINOv3 teacher. Given additional glimpses, CanViT-B reaches 45.9% ADE20K mIoU. On ImageNet-1k classification, CanViT-B also sets a new active-vision state of the art, with 84.5% top-1 accuracy after fine-tuning. CanViT generalizes to longer rollouts, larger scenes, and new policies. Our work narrows the wide gap between passive and active computer vision, demonstrating the potential of task- and policy-agnostic AVFM pretraining.

Checkpoints

The checkpoints are on Hugging Face under canvit; from_pretrained loads the PyTorch ones.

Checkpoint	Description
canvitb16-add-vpe-pretrain-g128px-s512px-in21k-dv3b16-2026-02-02	CanViT-B pretrained on ImageNet-21k by policy-agnostic passive-to-active dense latent distillation from DINOv3 ViT-B: the paper's model
canvitb16-add-vpe-pretrain-g128px-s512px-in1k-dv3b16-2026-06-22	CanViT-B pretrained the same way on ImageNet-1k only
canvitb16-add-vpe-finetune-g128px-s512px-in1k-2026-04-06	The ImageNet-21k model fine-tuned for ImageNet-1k classification, 84.5% top-1 (trained on TPU v6e via torch_xla)
canvitb16-add-vpe-finetune-g128px-s512px-in1k-2026-07-24	The ImageNet-21k model fine-tuned again with the same recipe, by a JAX/Flax NNX trainer
canvitb16-add-vpe-finetune-g128px-s512px-in1k-from-in1k-2026-07-24	The ImageNet-1k model fine-tuned with the same recipe, by the same trainer

Collections:

• CanViT ADE20K segmentation probes and DINOv3 ADE20K segmentation probes
• DINOv3 ImageNet-1k probes, which the frozen classification path uses
• CanViT pretraining ablations, the paper's ablation backbones
• JAX / Flax NNX and MLX copies, for the ports under Repository layout

Quickstart

We recommend uv for dependency management.

uv add canvit-pytorch   # or: pip install canvit-pytorch

from canvit_pytorch import CanViTForPretraining, Viewpoint, sample_at_viewpoint
from canvit_pytorch.preprocess import preprocess
from PIL import Image
import torch

# CanViT is integrated with the HuggingFace Hub.
model = CanViTForPretraining.from_pretrained(
    "canvit/canvitb16-add-vpe-pretrain-g128px-s512px-in21k-dv3b16-2026-02-02"
).eval()

# Replace with the image of your choice
image = Image.open("canvit-pytorch/test_data/Cat03.jpg").convert("RGB")
image = preprocess(512)(image)
image = image.unsqueeze(0)  # [1, 3, 512, 512]

# CanViT is a recurrent model.
state = model.init_state(batch_size=1, canvas_grid_size=32)

# Let's process a first glimpse: centered, zoomed-out.
# You can use any viewpoint you like, as long as it is within bounds.
# CanViT was trained on viewpoints covering 0.25% to 100%
# of a scene's surface area.
with torch.inference_mode():
    vp = Viewpoint.full_scene(batch_size=1, device=image.device)
    glimpse = sample_at_viewpoint(spatial=image, viewpoint=vp, glimpse_size_px=128)
    out = model(glimpse=glimpse, state=state, viewpoint=vp)

# Let's inspect the structure of what we get back.
# The canvas contains the model's working understanding of
# the scene at any given time, and is linearly decodable 
# into dense predictions upon token-wise LayerNorm.
# See `demos/basic.py` for how to visualize the canvas.
canvas_spatial = model.canvit.canvas_patch_grid(out.state.canvas)  # [1, 32, 32, 1024] — spatial feature map
out.state.recurrent_cls  # [1, 1, 768] — global CLS token
out.glimpse_patches      # [1, 64, 768] — glimpse patch features

# Now let's do a second glimpse: zoom into the top-left quadrant
# You can do this repeatedly: CanViT is recurrent with a large but constant-size canvas.
with torch.inference_mode():
    vp2 = Viewpoint(centers=torch.tensor([[-.5, -.5]]), scales=torch.tensor([.5]))
    glimpse2 = sample_at_viewpoint(spatial=image, viewpoint=vp2, glimpse_size_px=128)
    out2 = model(glimpse=glimpse2, state=out.state, viewpoint=vp2)
    
# You can use CanViT with frozen weights, fine-tune it, learn a policy on top...
# Or pretrain your own; it's fast.
# Start building!

ImageNet-1k Classification

CanViTForImageClassification provides a unified interface for classification. Two construction paths, same forward pass:

From a finetuned checkpoint (CanViT + head trained on IN1k):

from canvit_pytorch import CanViTForImageClassification, Viewpoint, sample_at_viewpoint
from canvit_pytorch.preprocess import preprocess
from PIL import Image
import torch

clf = CanViTForImageClassification.from_pretrained(
    "canvit/canvitb16-add-vpe-finetune-g128px-s512px-in1k-2026-04-06"
).eval()

From the frozen pretrained CanViT checkpoint + a DINOv3 linear probe:

clf = CanViTForImageClassification.from_pretrained_with_probe(
    pretrained_repo="canvit/canvitb16-add-vpe-pretrain-g128px-s512px-in21k-dv3b16-2026-02-02",
    probe_repo="canvit/dinov3-vitb16-lvd1689m-in1k-512x512-linear-clf-probe",
).eval()

Both have the same forward pass:

image = preprocess(512)(Image.open("canvit-pytorch/test_data/Cat03.jpg").convert("RGB")).unsqueeze(0)
state = clf.init_state(batch_size=1, canvas_grid_size=32)

with torch.inference_mode():
    vp = Viewpoint.full_scene(batch_size=1, device=image.device)
    glimpse = sample_at_viewpoint(spatial=image, viewpoint=vp, glimpse_size_px=128)
    logits, state = clf(glimpse=glimpse, state=state, viewpoint=vp)

print(logits.argmax(dim=-1))  # ImageNet-1k class index

ADE20K Semantic Segmentation

CanViTForSemanticSegmentation bundles a CanViT and a SegmentationProbe head into one model. forward returns per-pixel logits at canvas-grid resolution; predict adds bilinear upsampling.

from canvit_pytorch import CanViTForSemanticSegmentation

# Frozen CanViT + the flagship ADE20K probe (45.9% mIoU, 512px / 64x64 canvas):
seg = CanViTForSemanticSegmentation.from_pretrained_with_probe(
    pretrained_repo="canvit/canvitb16-add-vpe-pretrain-g128px-s512px-in21k-dv3b16-2026-02-02",
    probe_repo="canvit/probe-ade20k-40k-s512-c64-in21k",
).eval()

state = seg.init_state(batch_size=1, canvas_grid_size=64)
with torch.inference_mode():
    logits, state = seg(glimpse=glimpse, state=state, viewpoint=vp)               # [B, n_cls, 64, 64]
    upsampled, state = seg.predict(glimpse=glimpse, state=state, viewpoint=vp,
                                   target_size=(1024, 1024))                       # [B, n_cls, 1024, 1024]

The standalone SegmentationProbe head is also exported from canvit_pytorch for use on any spatial feature map. The published probes are listed under Checkpoints.

Demos

git clone https://github.com/m2b3/CanViT.git
cd CanViT/canvit-pytorch

# Classification with sequential glimpses
uv run --extra demo python demos/classify.py                     # finetuned checkpoint
uv run --extra demo python demos/classify.py --classifier frozen  # frozen CanViT + fused probe

# Canvas PCA visualization with two viewing strategies
uv run --extra demo python demos/basic.py

Supported platforms

• CPU
• CUDA (tested on RTX 4090, H100 SXM 80GB)
• Apple Silicon via MPS (inference, tested with torch 2.14)
• TPU via torch_xla 2.9.0 (tested on TPU v6e)

We aim to maintain compatibility with torch.export and ONNX Runtime. Please file an issue if you encounter problems.

Repository layout

canvit-pytorch/ holds the package:

Module	Contents	Docs
canvit_pytorch	The model: architecture, checkpoint loading, viewpoints and policies, task heads	this README
canvit_pytorch.flops	Analytic forward FLOPs of CanViT and DINOv3	docs/flops.md
canvit_pytorch.pretrain	Pretraining: passive-to-active dense distillation from DINOv3	docs/pretrain.md
canvit_pytorch.specialize	Downstream training: ADE20K segmentation probes, ImageNet-1k fine-tuning	docs/specialize.md
canvit_pytorch.evaluate	Evaluation and benchmarking: ADE20K mIoU, ImageNet-1k top-k, DINOv3 reconstruction	docs/evaluate.md
canvit_pytorch.viz	Recorded rollouts and smooth viewpoint paths for the project page and slides	docs/viz.md

site/ is the project page.

Pretraining, probe training and evaluation lived in separate repositories (CanViT-pretrain, CanViT-specialize, CanViT-eval), now archived; their histories are merged here.

Related repositories:

• dinov3-in1k-probes — ImageNet-1k linear probes for the DINOv3 ViTs, used by CanViT's classification path
• CanViT-MLX — MLX implementation for Apple Silicon (experimental)
• CanViT-NNX — JAX/Flax NNX implementation (experimental)

Troubleshooting

If you encounter errors loading pretrained checkpoints, ensure you are using the latest version of the package:

uv lock --upgrade-package canvit-pytorch && uv sync

Code written for canvit-pytorch 0.1 needs canvit-pytorch<0.2 and the checkpoints' earlier files: from_pretrained(..., revision="canvit-pytorch-0.1").

Citation

If you use this work, please cite our paper:

@article{berreby2026canvit,
  title={CanViT: Toward Active-Vision Foundation Models},
  author={Berreby, Yoha{\"i}-Eliel and Du, Sabrina and Durand, Audrey and Krishna, B. Suresh},
  year={2026},
  eprint={2603.22570},
  archivePrefix={arXiv},
  primaryClass={cs.CV},
  url={https://arxiv.org/abs/2603.22570}
}

Contact

Open an issue in this repository or email me@yberreby.com.

License

MIT. See LICENSE for details.