Tomato Lateral Shoot Images for Instance Segmentation in Greenhouses

A tomato lateral shoot image dataset was constructed using RGB imaging in greenhouse environments. The dataset was used to develop a YOLOv8n instance segmentation model with a Convolutional Block Attention Module, achieving a mAP of 98.1%. The dataset was created by Yingchun Jiang and last updated in June 2026.

Use Cases

• Training instance segmentation models for tomato lateral shoot detection based on the described RGB image collection.
• Evaluating model robustness under complex illumination and occlusion based on the described data augmentation strategies.
• Developing pruning decision support systems based on the integration of vision-based detection and LLM-assisted reasoning described in the framework.

Strengths

• Model trained on this data achieved a mAP of 98.1% for shoot detection.
• The associated YOLOv8n-seg+CBAM model has only 3.28M parameters and an average inference time of 8.0 ms per image.
• Data augmentation strategies were applied to simulate illumination variations, as described.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• The dataset is small, at 5.8 MB, which may limit the scale of model training.

Provenance

Source

figshare

Collection Method

Constructed using RGB imaging in greenhouse environments.

Freshness

Last updated 2026-06-02 05:40:18; freshness should be verified.

Geography

Greenhouse environments (specific location unknown).