turhancan97/visbench-keypoints2d-dinov2_vitb14

keypoints2d probe for dinov2_vitb14

A trained probe head, not a backbone. It is the small module VisBench fits on top of frozen dinov2_vitb14 features to measure what those features carry.

import visbench
from visbench.hub import load_probe_from_hub

backbone = visbench.get_backbone("dinov2_vitb14")
probe = load_probe_from_hub("turhancan97/visbench-keypoints2d-dinov2_vitb14", backbone=backbone)

It only works with this backbone

These weights were fitted on features from dinov2_vitb14, taken with pooling=mean and feature_mode=dense_only. Loading them against anything else is refused, because the failure is otherwise silent: a head fitted on one pooling and fed another has the right shapes and produces a plausible, wrong number.

backbone	dinov2_vitb14
backbone key	dinov2/dinov2_vitb14/224/7764ea0f912e
task	keypoints2d (low_level)
pooling	mean (requested mean)
feature mode	dense_only
layers	None

Reported scores

metric	value
keypoint_correlation	0.2248
mae	1.1294
rmse	2.5541

Reproducing it

Fitted with:

• activation: identity
• batch_size: 8
• epochs: 10
• head: linear
• hidden_dim: 512
• layers: None
• loss: l1
• lr: 0.0005
• optimizer: adamw
• protocol: visbench_keypoint2d_regression
• warmup_epochs: 1.5
• weight_decay: 0.0001

Generated by VisBench.