datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
4D-Animal
External Data for 4D-Animal
External data from the paper: 4D-Animal: Freely Reconstructing Animatable 3D Animals from Videos.
| ArXiv | Code |
Data Tree
.
├── cse # DensePose model weights and configuration files
│ ├── Base-DensePose-RCNN-FPN.yaml
│ ├── densepose_rcnn_R_50_FPN_soft_animals_I0_finetune_i2m_16k.yaml
│ └── model_final_8c9d99.pkl
├── lbos # Laplace-Beltrami Operator results
│ ├── lbo_cse.pk
│ ├── lbo_cse_to_smal.pk… See the full description on the dataset page: https://huggingface.co/datasets/zhongshsh/4D-Animal.animalspeak-pseudovox
AnimalSpeak Pseudovox Train-Unseen
This dataset contains the train-unseen split of AnimalSpeak Pseudovox. Each
example is a short, silence-trimmed, single-vocalization WAV clip plus compact
per-clip metadata. It does not include generated conversations, captions, QA
pairs, or MCQ answers.
Rows: 346,907
Shards: 18
Maximum rows per shard: 20,000
Files
data-20k/train-*.tar: WebDataset-style shards containing
audio/<audio_name> WAV entries.
metadata.parquet: one row per… See the full description on the dataset page: https://huggingface.co/datasets/EarthSpeciesProject/animalspeak-pseudovox.animal-180AnimalHaze3k
AnimalHaze3k Dataset
AnimalHaze3k is an image dehazing dataset focused on animal subjects in diverse environments. It contains 4,636 total images across training, validation, and test splits:
Training set: 1,041 clean reference images paired with 3,123 corresponding hazy versions
Test set: 59 clean images with 177 corresponding hazy versions
Validation set: 59 clean images with 177 corresponding hazy versions
wds_selected_animationanimal_activations