ModelListadosExterno

QTrack

QTrack is an end-to-end vision-language model designed for query-driven multi-object tracking (MOT). Unlike traditional MOT which tracks all objects in a scene, QTrack selectively localizes and tracks specific...

Fontehuggingface.co/GAASH-Lab/QTrack/tree/2d66b28d74d5b6aed816cb16f1986fb32c73c117

TipoOtherVersãoBf16 zero pp rank 0 mp rank 00 optim statesCnota

Fonte do modelo

Tipo: Other
Tamanho do arquivo: 14,3 KB–48,1 GB
Versão: vBf16 zero pp rank 0 mp rank 00 optim states
Atualizado: 24 de out. de 2026
Fonte: Hugging Face

Descrição da fonte

QTrack: Query-Driven Reasoning for Multi-modal MOT

QTrack is an end-to-end vision-language model designed for query-driven multi-object tracking (MOT). Unlike traditional MOT which tracks all objects in a scene, QTrack selectively localizes and tracks specific targets based on natural language instructions while maintaining temporal coherence and identity consistency.

Paper: QTrack: Query-Driven Reasoning for Multi-modal MOT
Project Page: https://gaash-lab.github.io/QTrack/
Repository: https://github.com/gaash-lab/QTrack

Description

Multi-object tracking has traditionally focused on estimating trajectories of all objects. QTrack introduces a query-driven tracking paradigm that formulates tracking as a spatiotemporal reasoning problem conditioned on natural language queries.

Key Contributions

RMOT26 Benchmark: A large-scale benchmark with grounded queries and sequence-level splits to enable robust evaluation of generalization.
QTrack Model: An end-to-end vision-language model that integrates multimodal reasoning with tracking-oriented localization.
Temporal Perception-Aware Policy Optimization (TPA-PO): A structured reward strategy to encourage motion-aware reasoning.

Benchmark Results

QTrack achieves state-of-the-art performance on the RMOT26 benchmark.

Model	Params	MCP↑	MOTP↑	CLE (px)↓	NDE↓
GPT-5.2	-	0.25	0.61	94.2	0.55
Qwen3-VL-Instruct	8B	0.25	0.64	96.0	0.97
Gemma 3	27B	0.24	0.56	58.4	0.88
InternVL	8B	0.21	0.66	117.44	0.64
QTrack (Ours)	3B	0.30	0.75	44.61	0.39

Installation

To set up the environment and use the model, please follow the instructions in the official repository:

# Create conda environment
conda create -n qtrack python=3.12
conda activate qtrack

# Install QTrack and dependencies
git clone https://github.com/gaash-lab/QTrack.git
cd QTrack
pip install -r requirements.txt
pip install -e .

Citation

If you find QTrack useful for your research, please cite:

@article{ashraf2026qtrack,
  title={QTrack: Query-Driven Reasoning for Multi-modal MOT},
  author={Ashraf, Tajamul and Tariq, Tavaheed and Yadav, Sonia and Ul Riyaz, Abrar and Tak, Wasif and Abdar, Moloud and Bashir, Janibul},
  journal={arXiv preprint arXiv:2603.13759},
  year={2026}
}

Fontes

1 fonte

Hugging FacePrimárioVersão 2d66b28d74d5b6aed816cb16f1986fb32c73c117

huggingface.co/GAASH-Lab/QTrack/tree/2d66b28d74d5b6aed816cb16f1986fb32c73c117

Verificado 24 de out.

Download files

Artefatos de modelo

36 artefatos

gemma3_finetunned/global_step2000/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt

pt · 48,1 GB · SHA-256 3f732233394a…18de · Hugging Face

Download file

gemma3_finetunned/global_step2000/zero_pp_rank_0_mp_rank_00_model_states.pt

pt · 471 KB · SHA-256 8da33f6c142e…6378 · Hugging Face

Download file

gemma3_finetunned/model-00001-of-00002.safetensors

safetensors · 4,62 GB · SHA-256 89425a5f971b…dd36 · Hugging Face

Download shard

gemma3_finetunned/model-00002-of-00002.safetensors

safetensors · 3,39 GB · SHA-256 c057c117d3fa…e8ab · Hugging Face

Download shard

MCP_Reward/actor/extra_state_world_size_1_rank_0.pt

pt · 14,3 KB · SHA-256 f7621bd46ae2…1610 · Hugging Face

Download file

MCP_Reward/actor/huggingface/model-00001-of-00004.safetensors

safetensors · 4,63 GB · SHA-256 61e51d920f6e…3e36 · Hugging Face

Download shard

MCP_Reward/actor/huggingface/model-00002-of-00004.safetensors

safetensors · 4,59 GB · SHA-256 997329cbe548…f37c · Hugging Face

Download shard

MCP_Reward/actor/huggingface/model-00003-of-00004.safetensors

safetensors · 4,59 GB · SHA-256 9badcc50723d…e858 · Hugging Face

Download shard

MCP_Reward/actor/huggingface/model-00004-of-00004.safetensors

safetensors · 1,33 GB · SHA-256 1294fb840efe…5fc8 · Hugging Face

Download shard

MCP_Reward/actor/model_world_size_1_rank_0.pt

pt · 15,1 GB · SHA-256 d6f232984be1…aeb7 · Hugging Face

Download file

MCP_Reward/actor/optim_world_size_1_rank_0.pt

pt · 28,0 GB · SHA-256 b7c016fef7cd…f4a7 · Hugging Face

Download file

QTrack/actor/extra_state_world_size_1_rank_0.pt

pt · 14,3 KB · SHA-256 909056981007…9282 · Hugging Face

Download file

QTrack/actor/huggingface/model-00001-of-00004.safetensors

safetensors · 4,63 GB · SHA-256 fa46b82c1f80…78b5 · Hugging Face

Download shard

QTrack/actor/huggingface/model-00002-of-00004.safetensors

safetensors · 4,59 GB · SHA-256 027f1c724b36…7611 · Hugging Face

Download shard

QTrack/actor/huggingface/model-00003-of-00004.safetensors

safetensors · 4,59 GB · SHA-256 3c3404a90058…62b8 · Hugging Face

Download shard

QTrack/actor/huggingface/model-00004-of-00004.safetensors

safetensors · 1,33 GB · SHA-256 9720b510f279…f352 · Hugging Face

Download shard

QTrack/actor/model_world_size_1_rank_0.pt

pt · 15,1 GB · SHA-256 f291f9fb3e3e…3fff · Hugging Face

Download file

QTrack/actor/optim_world_size_1_rank_0.pt

pt · 28,0 GB · SHA-256 7caadba0e7b2…ab4d · Hugging Face

Download file

Qwen2.5_vl_finetune/global_step6800/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt

pt · 34,9 GB · SHA-256 a147a58f37f1…53f3 · Hugging Face

Download file

Qwen2.5_vl_finetune/global_step6800/zero_pp_rank_0_mp_rank_00_model_states.pt

pt · 1,18 GB · SHA-256 00441a8d1ab4…3c2b · Hugging Face

Download file

Qwen2.5_vl_finetune/model-00001-of-00002.safetensors

safetensors · 4,65 GB · SHA-256 92912fa64675…331c · Hugging Face

Download shard

Qwen2.5_vl_finetune/model-00002-of-00002.safetensors

safetensors · 2,34 GB · SHA-256 5e24a3d7bda1…6a7c · Hugging Face

Download shard

TAPO_Only/actor/extra_state_world_size_1_rank_0.pt

pt · 14,3 KB · SHA-256 e7777b21f51c…fbda · Hugging Face

Download file

TAPO_Only/actor/huggingface/model-00001-of-00004.safetensors

safetensors · 4,63 GB · SHA-256 8ba9868aa8b9…1410 · Hugging Face

Download shard

TAPO_Only/actor/huggingface/model-00002-of-00004.safetensors

safetensors · 4,59 GB · SHA-256 e6d197c36ea7…8670 · Hugging Face

Download shard

TAPO_Only/actor/huggingface/model-00003-of-00004.safetensors

safetensors · 4,59 GB · SHA-256 0236c8bd535b…0692 · Hugging Face

Download shard

TAPO_Only/actor/huggingface/model-00004-of-00004.safetensors

safetensors · 1,33 GB · SHA-256 ffada86db25d…56fc · Hugging Face

Download shard

TAPO_Only/actor/model_world_size_1_rank_0.pt

pt · 15,1 GB · SHA-256 fb63ac27e9d8…6b4f · Hugging Face

Download file

TAPO_Only/actor/optim_world_size_1_rank_0.pt

pt · 28,0 GB · SHA-256 46ab14a3c7cb…ec57 · Hugging Face

Download file

visionReasoner_finetunned/actor/extra_state_world_size_1_rank_0.pt

pt · 14,3 KB · SHA-256 909056981007…9282 · Hugging Face

Download file

visionReasoner_finetunned/actor/huggingface/model-00001-of-00004.safetensors

safetensors · 4,63 GB · SHA-256 f961f072b39a…14ad · Hugging Face

Download shard

visionReasoner_finetunned/actor/huggingface/model-00002-of-00004.safetensors

safetensors · 4,59 GB · SHA-256 2f4efcbb3e43…62aa · Hugging Face

Download shard

visionReasoner_finetunned/actor/huggingface/model-00003-of-00004.safetensors

safetensors · 4,59 GB · SHA-256 e73c91f6e6cd…0a71 · Hugging Face

Download shard

visionReasoner_finetunned/actor/huggingface/model-00004-of-00004.safetensors

safetensors · 1,33 GB · SHA-256 7a9ff903964c…1914 · Hugging Face

Download shard

visionReasoner_finetunned/actor/model_world_size_1_rank_0.pt

pt · 15,1 GB · SHA-256 271eed4901fd…9647 · Hugging Face

Download file

visionReasoner_finetunned/actor/optim_world_size_1_rank_0.pt

pt · 28,0 GB · SHA-256 ccf7305066f3…953b · Hugging Face

Download file

Trechos de fonte

2 trechos

Readme6 de ago.huggingface.co/GAASH-Lab/QTrack

--- license: mit pipeline_tag: video-text-to-text tags: - multi-object-tracking - video-understanding - vision-language-model - spatiotemporal-reasoning --- # QTrack: Query-Driven Reasoning for Multi-modal MOT [**QTrack**](https://huggingface.co/papers/2603.13759) is an end-to-end vision-language model designed for query-driven multi-object tracking (MOT). Unlike traditional MOT which tracks all objects in a scene, QTrack selectively localizes and tracks specific targets based on natural language instructions while maintaining temporal coherence and identity consistency. - **Paper:** [QTrack: Query-Driven Reasoning for Multi-modal MOT](https://huggingface.co/papers/2603.13759) - **Project Page:** [https://gaash-lab.github.io/QTrack/](https://gaash-lab.github.io/QTrack/) - **Repository:** [https://github.com/gaash-lab/QTrack](https://github.com/gaash-lab/QTrack) ## Description Multi-object tracking has traditionally focused on estimating trajectories of all objects. QTrack introduces a **query-driven tracking paradigm** that formulates tracking as a spatiotemporal reasoning problem conditioned on natural language queries. ### Key Contributions - **RMOT26 Benchmark**: A large-scale benchmark with grounded queries and sequence-level splits to enable robust evaluation of generalization. - **QTrack Model**: An end-to-end vision-language model that integrates multimodal reasoning with tracking-oriented localization. - **Temporal Perception-Aware Policy Optimization (TPA-PO)**: A structured reward strategy to encourage motion-aware reasoning. ## Benchmark Results QTrack achieves state-of-the-art performance on the [RMOT26](https://huggingface.co/datasets/GAASH-Lab/RMOT26) benchmark. | Model | Params | MCP↑ | MOTP↑ | CLE (px)↓ | NDE↓ | |:-----:|:------:|:----:|:-----:|:---------:|:----:| | GPT-5.2 | - | 0.25 | 0.61 | 94.2 | 0.55 | | Qwen3-VL-Instruct | 8B | 0.25 | 0.64 | 96.0 | 0.97 | | Gemma 3 | 27B | 0.24 | 0.56 | 58.4 | 0.88 | | InternVL | 8B | 0.21 | 0.66 | 117.44 | 0.64 | | **QTrack (Ours)** | **3B** | **0.30** | **0.75** | **44.61** | **0.39** | ## Installation To set up the environment and use the model, please follow the instructions in the [official repository](https://github.com/gaash-lab/QTrack): ```bash # Create conda environment conda create -n qtrack python=3.12 conda activate qtrack # Install QTrack and dependencies git clone https://g...

Note6 de ago.huggingface.co/GAASH-Lab/QTrack

Source context: 0 downloads · 0 likes · Pipeline video-text-to-text · Repo GAASH-Lab/QTrack