Downloads · 30 days
0
TasselNetV4
akaxiao/TasselNetV4
TasselNetV4 is a machine learning model from akaxiao. Use it for the machine learning task on the model card, and read the license before you ship it in a product. The card lists the license as apache-2.0.
Official implementation of [TasselNetV4: A vision foundation model for cross-scene, cross-scale, and cross-species plant counting]
Access
Public
Updated Oct 9, 2025
Repo size
1.1 GB
Likes
0
Public
Repo makeup
Click a slice to open those files.
.pth1.1 GB · 98%
At a glance
- License
- apache-2.0
- Access
- Public
- Created
- Sep 29, 2025
- Updated
- Oct 9, 2025
- SHA
- b469b541
Model card
From the Hugging Face model README
TasselNetv4
Official implementation of [TasselNetV4: A vision foundation model for cross-scene, cross-scale, and cross-species plant counting]
Paper link: https://huggingface.co/papers/2509.20857
Highlights
- Plant agnostic counting: a new plant-orientated task customizing Class-agnostic counting into the plant domain and highlighting zero-shot generalization across taxomomic plant species;
- PAC-105 and PAC-Somalia: two challenging PAC datasets for training and evaluating daily and out-of-distribution plant species;
- TasselNetV4: an extended version of the TasselNet plant counting models.
Comparison with the state-of-the-art CAC approaches on the PAC-105 dataset. Best performance is in boldface.
| Method | Venue & Year | Shot | MAE↓ | RMSE↓ | WCA↑ | $R^2$ ↑ | MPE |↓| | FPS |
|---|---|---|---|---|---|---|---|---|
| FamNet (Ranjan et al., 2021) | CVPR'21 | 3 | 31.70 | 62.58 | 0.49 | 0.56 | 0.24 | 89.65 |
| BMNet+ (Shi et al., 2022) | CVPR'22 | 3 | 27.03 | 60.18 | 0.56 | 0.61 | 0.02 | 43.08 |
| SPDCNet (Lin et al., 2022) | BMVC'22 | 3 | 25.21 | 49.98 | 0.58 | 0.92 | 0.21 | 155.84 |
| SAFECount (You et al., 2023) | WACV'23 | 3 | 25.59 | 52.09 | 0.58 | 0.91 | 0.04 | 118.55 |
| CountTR (Liu et al., 2022) | BMVC'22 | 3 | 25.25 | 49.31 | 0.63 | 0.92 | 0.12 | 112.46 |
| T-Rex2 (Jiang et al., 2025) | ECCV'24 | 3 | 26.04 | 49.31 | 0.58 | 0.92 | -0.13 | \ |
| CACViT (Wang et al., 2024b) | AAAI'24 | 3 | 19.51 | 29.59 | 0.68 | 0.89 | 0.29 | 89.65 |
| TasselNetV4 (Ours) | This Paper | 3 | 16.04 | 28.03 | 0.74 | 0.92 | -0.05 | 121.62 |
| FamNet (Ranjan et al., 2021) | CVPR'21 | 1 | 35.91±0.966 | 71.78±1.188 | 0.42±0.014 | 0.45±0.024 | 0.57±0.017 | \ |
| BMNet (Shi et al., 2022) | CVPR'22 | 1 | 28.78±0.324 | 62.12±0.437 | 0.15±0.001 | 0.59±0.008 | 0.13±0.006 | \ |
| CountTR (Liu et al., 2022) | BMVC'22 | 1 | 28.46±0.226 | 49.84±0.646 | 0.70±0.037 | 0.73±0.006 | 0.11±0.003 | \ |
| CACViT (Wang et al., 2024b) | AAAI'24 | 1 | 21.80±0.429 | 38.40±1.526 | 0.64±0.005 | 0.84±0.013 | 0.29±0.002 | \ |
| TasselNetV4 (Ours) | This Paper | 1 | 18.04±0.339 | 32.04±1.213 | 0.71±0.005 | 0.90±0.009 | 0.02±0.000 | \ |
Comparison with the state-of-the-art CAC approaches on the PAC-Somalia dataset. Best performance is in boldface.
| Method | Venue & Year | Shot | MAE↓ | RMSE↓ | WCA↑ | $R^2$ ↑ | MPE |↓| |
|---|---|---|---|---|---|---|---|
| CountTR (Liu et al., 2022) | BMVC'22 | 3 | 12.71 | 23.87 | 0.38 | 0.57 | 0.27 |
| CACViT (Wang et al., 2024b) | AAAI'24 | 3 | 14.00 | 17.00 | 0.55 | 0.78 | 0.80 |
| TasselNetV4 (Ours) | This Paper | 3 | 8.88 | 13.11 | 0.72 | 0.87 | 0.32 |
| CountTR (Liu et al., 2022) | BMVC'22 | 1 | 12.79±0.076 | 24.20±0.292 | 0.37±0.005 | 0.55±0.012 | 0.23±0.062 |
| CACViT (Wang et al., 2024b) | AAAI'24 | 1 | 14.74±0.138 | 18.23±0.446 | 0.53±0.005 | 0.75±0.012 | 0.80±0.008 |
| TasselNetV4 (Ours) | This Paper | 1 | 10.98±0.065 | 16.73±0.150 | 0.65±0.000 | 0.80±0.005 | 0.42±0.001 |
Visualization
- License
- apache-2.0
- Created
- Sep 29, 2025
- Updated
- Oct 9, 2025