---
license: cc-by-nc-sa-4.0
task_categories:
  - visual-question-answering
  - image-classification
  - image-to-text
language:
  - en
tags:
  - dental
  - medical-imaging
  - vlm-benchmark
  - panoramic-radiograph
  - intraoral-photograph
  - periapical-radiograph
  - cephalometric-radiograph
  - lora
  - on-device-deployment
size_categories:
  - 10K<n<100K
pretty_name: "Pocket-Dentist Benchmark"
viewer: false
configs:
  - config_name: default
    data_files:
      - split: train
        path: "data/**"
---

# Pocket-Dentist-Bench

<!-- **Pocket-Dentist: Benchmarking Compact Vision-Clinical Disclaimer:Language Models for Dental Image Understanding** -->

<p align="center"><span style="color: red;">🚧 Dataset Coming Soon — The full benchmark data will be released upon paper acceptance. </span> <br/> Please check back later or ⭐ star this repo to get notified.</p>

---

## Links

- 📄 **Paper**: Coming soon
- 💻 **Code**: [GitHub](https://pocket-dentist.github.io/Pocket-Dentist-Bench?utm_source=wechat&utm_medium=social&utm_medium=referral)
- 🤗 **Models**: Coming soon

---

## Overview

Pocket-Dentist is a large-scale multimodal benchmark and deployment pipeline for evaluating Vision-Language Models (VLMs) on dental image understanding tasks. The benchmark curates and standardizes **seven dental datasets** into a unified vision-language evaluation framework.

**Key Statistics:**
- 🏥 **6,000+** patients
- 🖼️ **71,000+** images
- 📷 **4** imaging modalities
- 📋 **6** task types
- 📊 **14** evaluation metrics
- 🤖 **14** VLMs benchmarked (including 12 open-weight models with LoRA adaptation)

<!-- <p align="center">
  <img src="assets/data_overview.png" alt="Pocket-Dentist Benchmark Overview" width="100%">
</p>

<p align="center"><em>Figure 1: Overview of the Pocket-Dentist benchmark datasets and task formulations. Seven dental datasets are unified across four imaging modalities, 14 evaluation tasks, six task types, and 14 evaluation metrics.</em></p> -->

---

## Benchmark Pipeline

The Pocket-Dentist evaluation pipeline consists of four stages:

1. **Data Collection & Unification** — Curating heterogeneous dental datasets into a unified multimodal benchmark
2. **Task Design & Annotation** — Converting source annotations into shared prompt–response task formats
3. **Model Evaluation & Adaptation** — Evaluating VLMs under zero-shot, few-shot, and LoRA adaptation settings
4. **On-Device Deployment** — Measuring local inference efficiency for compact adapted models on mobile hardware

<p align="center">
  <img src="assets/pipeline.png" alt="Pocket-Dentist Pipeline" width="100%">
</p>

<p align="center"><em>Figure 1: Deploy-aware evaluation pipeline of Pocket-Dentist.</em></p>

---

## Benchmark Results

### Zero-Shot Performance

Under zero-shot evaluation, no single model dominates across all 14 metrics. Closed-source APIs (Gemini) perform best overall, while compact model performance is fragmented across tasks.

<details>
<summary><b>Table 1: Zero-Shot (ZS) Results — Click to expand</b></summary>

**Large VLMs (≥7B)**

| Model | BRAR Acc | BRAR F1 | DR F1w | Meta VQA | Meta Cap | Meta Cls | Aariz VQA | Aariz CVM | COde Cls | DenPAR Arch | DenPAR Site | DenPAR MAE↓ | Caries Det | Caries Cls |
|-------|---------|---------|--------|----------|----------|----------|-----------|-----------|----------|-------------|-------------|------------|------------|------------|
| Lingshu-32B | 0.49 | **0.39** | 0.60 | 0.63 | 0.18 | 0.34 | 0.26 | 0.13 | 0.48 | 0.59 | **0.53** | 0.88 | 0.56 | 0.16 |
| MedMO-8B-Next | 0.26 | 0.19 | 0.53 | 0.49 | 0.09 | 0.08 | 0.21 | 0.05 | 0.26 | 0.61 | 0.29 | 2.90 | 0.59 | 0.84 |
| Qwen2.5-VL-7B | 0.27 | 0.17 | 0.32 | 0.45 | 0.15 | 0.23 | 0.20 | 0.00 | 0.50 | 0.40 | 0.35 | 1.01 | 0.63 | 0.14 |
| gemini-2.0-flash | **0.57** | 0.37 | 0.00 | 0.63 | 0.18 | **0.36** | 0.29 | **0.25** | 0.54 | 0.84 | 0.45 | **0.42** | 0.50 | 0.12 |
| gemini-2.5-flash | 0.27 | 0.26 | **0.62** | 0.66 | 0.14 | 0.24 | 0.23 | 0.12 | **0.58** | **0.99** | 0.51 | 0.47 | 0.54 | 0.13 |

**Compact VLMs (≤4B)**

| Model | BRAR Acc | BRAR F1 | DR F1w | Meta VQA | Meta Cap | Meta Cls | Aariz VQA | Aariz CVM | COde Cls | DenPAR Arch | DenPAR Site | DenPAR MAE↓ | Caries Det | Caries Cls |
|-------|---------|---------|--------|----------|----------|----------|-----------|-----------|----------|-------------|-------------|------------|------------|------------|
| Qwen3.5-4B | 0.17 | 0.10 | 0.54 | **0.82** | 0.10 | 0.16 | 0.17 | 0.04 | 0.11 | 0.40 | 0.19 | 3.02 | 0.49 | 0.18 |
| Qwen3-VL-4B | 0.44 | 0.37 | 0.24 | 0.58 | **0.20** | 0.22 | 0.23 | 0.08 | 0.54 | 0.44 | 0.23 | 0.42 | 0.63 | 0.58 |
| gemma-4-E4B-it | 0.56 | 0.24 | 0.61 | 0.59 | 0.18 | 0.31 | 0.31 | 0.04 | 0.51 | 0.40 | 0.51 | 0.52 | 0.43 | 0.30 |
| medgemma-4b-it | 0.44 | 0.33 | 0.57 | 0.54 | 0.14 | 0.16 | **0.40** | 0.03 | 0.27 | 0.40 | 0.23 | 0.89 | 0.52 | 0.11 |
| paligemma2-3b | 0.10 | 0.06 | 0.00 | 0.00 | 0.00 | 0.00 | 0.20 | 0.03 | 0.00 | 0.00 | 0.18 | 0.89 | **0.64** | 0.00 |
| SmolVLM2-2.2B | 0.56 | 0.35 | 0.56 | 0.00 | 0.10 | 0.15 | 0.23 | 0.05 | 0.10 | 0.60 | 0.10 | 0.89 | 0.44 | **0.92** |
| InternVL3.5-2B | 0.50 | 0.27 | 0.09 | 0.15 | 0.00 | 0.00 | 0.37 | 0.00 | 0.14 | 0.40 | 0.22 | 3.21 | 0.36 | 0.12 |
| gemma-4-E2B-it | 0.56 | 0.24 | 0.24 | 0.48 | 0.15 | 0.25 | 0.39 | 0.03 | 0.50 | 0.27 | 0.23 | 0.73 | 0.61 | 0.11 |
| InternVL3.5-1B | 0.26 | 0.14 | 0.62 | 0.34 | 0.00 | 0.00 | 0.21 | 0.07 | 0.14 | 0.28 | 0.19 | 2.27 | 0.61 | 0.11 |

> **Bold** = best in tier. ↑ higher is better; MAE ↓ lower is better.

</details>

---

## On-Device Deployment

We deploy LoRA-tuned VLMs on an **iPhone 17 Pro** (A19 Pro SoC, 12 GB Unified Memory) via Metal-accelerated inference using `llama.cpp`. All computation is performed locally on the device with 100% offline privacy protection.

| Model | Total Latency (s) ↓ | TTFT (s) ↓ | ITPS (t/s) ↑ | OTPS (t/s) ↑ | RAM (GB) ↓ |
|-------|---------------------|------------|--------------|--------------|------------|
| **Pocket-Dentist-4B** | 6.67 | 1.22 | 315.95 | 17.07 | 4.09 |
| InternVL3.5-2B | **4.74** | **0.78** | **434.53** | **29.47** | **2.62** |
| Qwen2.5-VL-7B | 24.06 | 2.29 | 148.93 | 9.60 | 6.22 |

<p align="center">
  <img src="assets/deployment.png" alt="Pocket-Dentist iOS App" width="280">
</p>

<p align="center"><em>Pocket-Dentist iOS app running Pocket-Dentist-4B locally on an iPhone 17 Pro.</em></p>

---

## Key Findings

- 🔍 **Zero-shot fragmentation**: No single model dominates across all dental tasks under zero-shot evaluation
- 📈 **LoRA adaptation closes the gap**: Under a uniform LoRA budget, compact VLMs become competitive with substantially larger models
- 🏆 **Qwen3-VL-4B achieves the strongest overall performance** among compact models, matching or outperforming larger open-weight models (7B–32B) on most primary task metrics
- 📱 **Pocket-Dentist-4B** (LoRA-tuned Qwen3-VL-4B) runs locally on an iPhone 17 Pro with **6.67s** per-sample latency and **4.09 GB** RAM
- 🏥 Medical pre-training alone does not guarantee dental task performance — dental-domain LoRA adaptation is more effective

---

## License & Disclaimer

This benchmark integrates data from multiple publicly available dental imaging datasets, each with its own license. This repository is distributed under [CC BY-NC-SA 4.0](https://creativecommons.org/licenses/by-nc-sa/4.0/). Users must also comply with the individual licenses of the constituent datasets.

---

<!--
## Citation

```bibtex
@article{pocket-dentist-2026,
  title={Pocket-Dentist: Benchmarking Compact Vision-Language Models for Dental Image Understanding},
  author={TODO},
  journal={arXiv preprint arXiv:26XX.XXXXX},
  year={2026}
}
```
-->