datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
coda-lm-llava-format
CODA-LM Dataset Card
CODA-LM is the multi-modal version of the CODA dataset, used in the CODA-LM paper. Both English and Chinese annotations are available. Check detailed usage in our Github repo.
This repo contains the CODA-LM dataset, which has been reorganized in the LLaVA data format.
You are also welcome to check the original CODA-LM data which contains more metadata vanilla annotations.
Usage
from datasets import load_dataset
# name can be selected from… See the full description on the dataset page: https://huggingface.co/datasets/KaiChen1998/coda-lm-llava-format.eurocv
EuroCV - images dataset
This dataset was created using codaco.app.
Description
Develop better AI together: Your images help build diverse training data, making AI systems more accurate, safer, and more reliable. Join now and help improve AI!
Labels
This dataset includes the following labels:
Bounding box objects
License
This dataset is licensed under CC BY 4.0.
You are free to share and adapt it for any purpose, including… See the full description on the dataset page: https://huggingface.co/datasets/codaco/eurocv.vlmn_tartandrive100_scand50_coda25_spot100_sub5_full_augmentation_processed_10
Trajectory Ranking Dataset
This dataset contains trajectory ranking results for autonomous navigation scenarios.
Dataset Statistics
Total examples: 39558
Chunks processed: 40
Upload date: 2025-09-13T00:44:30.335177
Features
Image data with terrain analysis
Trajectory rankings and reasoning
Quality and diversity analysis
Terrain and trajectory descriptions
images
CoDaCo - images dataset
This dataset was created using codaco.app.
Description
All data contributed to this campaign goes to the global CoDaCo datasets.
Labels
This dataset includes the following labels:
Captions
Bounding box objects
Bounding box texts
Contains objects
Tags
Emotions
AI generated
Quality rating
License
This dataset is licensed under CC BY 4.0.
You are free to share and adapt it for any purpose, including commercially… See the full description on the dataset page: https://huggingface.co/datasets/codaco/images.microscope-datavlmn_iphone100_tartandrive100_scand50_coda25_spot100_sub5
vlmn_iphone100_tartandrive100_scand50_coda25_spot100_sub5
Description
VLN Navigation dataset with 100% of iphone data, 100% of tartandrive data, 50% of scand data, 25% of coda data, and 100% of in-domain spot data. Whenever daatsets aren't 100%, they are ranked by curvature and output of length 5.
Processing Parameters
{}
Dataset Configuration
Train dataset:
mixer: mateoguaman/coda_every1_25pct_sub5: 1.0
mateoguaman/iphone_stairs_ramps: 1.0… See the full description on the dataset page: https://huggingface.co/datasets/mateoguaman/vlmn_iphone100_tartandrive100_scand50_coda25_spot100_sub5.coda_every1_25pct_sub5
coda_every1_25pct_sub5
Description
Processed coda dataset with filter_every_nth=1, 25% of data, and num_subsampled_points=5
Processing Parameters
mateoguaman/coda:
exclude_outliers_pct: 3
filter_by_curvature: true
filter_every_nth: 1
horizon:
300: 0.5
500: 0.5
num_subsampled_points: 5
Dataset Configuration
Train dataset:
mixer: mateoguaman/coda: 0.25
split: train
Validation dataset:
mixer: mateoguaman/coda: 0.25split:… See the full description on the dataset page: https://huggingface.co/datasets/mateoguaman/coda_every1_25pct_sub5.vlmn_tartandrive100_scand50_coda25_spot100_sub5_filtered_trajectories_training_25_fixedcodacoda
CODa Navigation Dataset
This dataset contains navigation trajectory data for robotic navigation tasks. Each example includes an RGB image, a language goal describing the desired navigation target, and 2D/3D trajectories showing the path to the goal.
Dataset Structure
image: RGB image from the robot's viewpoint
lang_goal: Natural language instruction describing the navigation goal
trajectory_2d: 2D trajectory coordinates (pixel space)
trajectory_3d: 3D trajectory… See the full description on the dataset page: https://huggingface.co/datasets/mateoguaman/coda.vlmn_iphone100_tartandrive100_scand50_coda25_spot100_insta360100_sub5
vlmn_iphone100_tartandrive100_scand50_coda25_spot100_insta360100_sub5
Description
VLN Navigation dataset with 100% of iphone data, 100% of tartandrive data, 50% of scand data, 25% of coda data, 100% of in-domain spot data, and 100% of insta360 data. Whenever daatsets aren't 100%, they are ranked by curvature and output of length 5.
Processing Parameters
{}
Dataset Configuration
Train dataset:
mixer: mateoguaman/coda_every1_25pct_sub5: 1.0… See the full description on the dataset page: https://huggingface.co/datasets/mateoguaman/vlmn_iphone100_tartandrive100_scand50_coda25_spot100_insta360100_sub5.coda_every1_25pct_rdp
coda_every1_25pct_rdp
Description
Processed coda dataset with filter_every_nth=1, 25% of data, and subsampled with RDP.
Processing Parameters
mateoguaman/coda:
exclude_outliers_pct: 3
filter_by_curvature: true
filter_every_nth: 1
horizon:
300: 0.5
500: 0.5
num_subsampled_points: -1
subsample_method: rdp
tolerance: 25
Dataset Configuration
Train dataset:
mixer: mateoguaman/coda: 0.25
split: train
Validation dataset:… See the full description on the dataset page: https://huggingface.co/datasets/mateoguaman/coda_every1_25pct_rdp.vlmn_tartandrive100_scand50_coda25_spot100_sub5_filtered_trajectories_training_10_fixedv0.4.1_codatasetvlmn_tartandrive100_scand50_coda25_spot100_sub5_filtered_trajectories_training_10_fixedvlmn_tartandrive100_scand50_coda25_spot100_insta360100_sub5
vlmn_tartandrive100_scand50_coda25_spot100_insta360100_sub5
Description
VLN Navigation dataset with 100% of tartandrive data, 50% of scand data, 25% of coda data, 100% of in-domain spot data, and 100% of insta360 data. Whenever daatsets aren't 100%, they are ranked by curvature and output of length 5.
Processing Parameters
{}
Dataset Configuration
Train dataset:
mixer: mateoguaman/coda_every1_25pct_sub5: 1.0
mateoguaman/insta360_every1_sub5: 1.0… See the full description on the dataset page: https://huggingface.co/datasets/mateoguaman/vlmn_tartandrive100_scand50_coda25_spot100_insta360100_sub5.vlmn_tartandrive100_scand50_coda25_spot100_sub5_filtered_trajectories_training_25_fixedvlmn_tartandrive100_scand50_coda25_spot100_sub5_full_augmentation_25CODA-LM_raw_dataSAVANT-CODALM-medium
SAVANT CODALM Medium Dataset
This dataset is part of the SAVANT framework described in the SAVANT paper, currently under peer review.
This repository is provided for peer-review purposes only. After the review process, the dataset will be made publicly available through the authors' main account.
Dataset Description
CODALM medium was created by combining automated framework evaluation with human validation. Starting with the full CODA dataset (9,640 images), we used… See the full description on the dataset page: https://huggingface.co/datasets/u94fmn391j/SAVANT-CODALM-medium.vlmn_iphonecf100_tartandrive100_scand50_coda25_spot100_rdp
vlmn_iphonecf100_tartandrive100_scand50_coda25_spot100_rdp
Description
VLN Navigation dataset with 100% of counterfactual iphone data, 100% of tartandrive data, 50% of scand data, 25% of coda data, and 100% of in-domain spot data. Whenever daatsets aren't 100%, they are ranked by curvature and output of length 5.
Processing Parameters
{}
Dataset Configuration
Train dataset:
mixer: mateoguaman/coda_every1_25pct_rdp: 1.0… See the full description on the dataset page: https://huggingface.co/datasets/mateoguaman/vlmn_iphonecf100_tartandrive100_scand50_coda25_spot100_rdp.CoDASAVANT-CODALM-large
SAVANT (Semantic Anomaly Verification/Analysis Toolkit)
Project Page | Paper
SAVANT is a model-agnostic reasoning framework that reformulates anomaly detection in autonomous driving as a layered semantic consistency verification. This dataset consists of approximately 10,000 real-world driving images curated to address the challenge of detecting rare, out-of-distribution semantic anomalies.
The dataset includes structured scene descriptions and multi-modal evaluations, which were… See the full description on the dataset page: https://huggingface.co/datasets/Brusnicki/SAVANT-CODALM-large.SAVANT-CODALM-small
SAVANT CODALM Small Dataset
This dataset is part of the SAVANT framework described in the SAVANT paper, currently under peer review.
This repository is provided for peer-review purposes only. After the review process, the dataset will be made publicly available through the authors' main account.
Dataset Description
CODALM small contains 100 real-world driving images (50 anomalous, 50 normal) derived from the CODA corner case dataset. Each image includes manual annotation… See the full description on the dataset page: https://huggingface.co/datasets/u94fmn391j/SAVANT-CODALM-small.vlmn_tartandrive100_scand50_coda25_spot100_rdp
vlmn_tartandrive100_scand50_coda25_spot100_rdp
Description
VLN Navigation dataset with 100% of tartandrive data, 50% of scand data, 25% of coda data, and 100% of in-domain spot data. Whenever daatsets aren't 100%, they are ranked by curvature and subsampled with RDP.
Processing Parameters
{}
Dataset Configuration
Train dataset:
mixer: mateoguaman/coda_every1_25pct_rdp: 1.0
mateoguaman/scand_every1_50pct_rdp: 1.0
mateoguaman/spot_every1_100pct_rdp:… See the full description on the dataset page: https://huggingface.co/datasets/mateoguaman/vlmn_tartandrive100_scand50_coda25_spot100_rdp.vlmn_tartandrive100_scand50_coda25_spot100_sub5_full_augmentationNovelSpecies-CoDA-SubsetSUN-CoDA-Subsetvlmn_tartandrive100_scand50_coda25_spot100_sub5_augmented_unfiltered_2SAVANT-CODALM-medium
SAVANT (Semantic Anomaly Verification/Analysis Toolkit)
Project Page | Paper
SAVANT is a model-agnostic reasoning framework designed to improve semantic anomaly detection in autonomous driving. It reformulates anomaly detection as a layered semantic consistency verification task. This dataset contains approximately 10,000 real-world images annotated with structured reasoning, used to fine-tune Vision-Language Models (VLMs) like Qwen2.5-VL for high-accuracy, single-shot anomaly… See the full description on the dataset page: https://huggingface.co/datasets/Brusnicki/SAVANT-CODALM-medium.