You can deploy this model and start earning money today!

HSP-IIT/
fsosar-strm-frdisc-hmdb51-5shot

Name: fsosar-strm-frdisc-hmdb51-5shot
Availability: InStock

Views

No views yet

STRM — FR-Disc — HMDB51 (5-shot)

Checkpoint from A Baseline Study and Benchmark for Few-Shot Open-Set Action Recognition with Feature Residual Discrimination (Berti, Pasquale, Natale — ICPR 2026), trained with the discriminator (FR-Disc) open-set loss.

Model: STRM
Dataset: HMDB51
Setting: 5-way 5-shot
Open-set method: FR-Disc (feature-residual discriminator)
Checkpoint eval metrics (measured for this exact checkpoint): FS-ACC = 75.50, OS-ACC = 56.87

This configuration is not part of the results table in the paper/README (STRM's FR-Disc results were not part of the final reported benchmark); this is simply the best available checkpoint found in our training logs for this setting.

Code, other checkpoints, and the FSOS-AR benchmark splits are at hsp-iit/fsosar.

Usage

python

from huggingface_hub import hf_hub_download
import torch

ckpt_path = hf_hub_download(repo_id="HSP-IIT/fsosar-strm-frdisc-hmdb51-5shot", filename="model.pt")
state_dict = torch.load(ckpt_path, map_location="cpu")

# Requires the model definition from https://github.com/hsp-iit/fsosar
from models import STRM
from utils import load_configs

config = load_configs("STRM", "HMDB51")
model = STRM(config, disc=True, gc=False, dp=config["dp"])
model.load_state_dict(state_dict, strict=True)
model.eval()

License

BSD 3-Clause (matching the code repository) applies to the model code/weights format. Note: this checkpoint was trained on HMDB51, which is licensed CC BY-NC-SA 3.0 (Non-Commercial). Use of this checkpoint should be limited to non-commercial research purposes, consistent with the source dataset's license. See the full audit at hsp-iit/workbook-stefano-berti#70.

Citation

bibtex

@inproceedings{berti2026fsosar,
  title     = {A Baseline Study and Benchmark for Few-Shot Open-Set Action Recognition with Feature Residual Discrimination},
  author    = {Berti, Stefano and Pasquale, Giulia and Natale, Lorenzo},
  booktitle = {International Conference on Pattern Recognition (ICPR)},
  year      = {2026}
}