GroMo25

LEADERBOARD

Age Prediction – MAE Scores

Rank	Team	Okra	Radish	Mustard	Wheat	Average
1	ETTI 2 - VIS	1.71	1.61	6.47	2.90	3.17
2	DeepLeaf	4.80	4.60	7.80	6.15	5.83
3	AIgriTech	3.77	5.03	8.70	8.44	6.48
4	PlantPixels	13.10	5.60	3.20	7.30	7.30
5	Resense	12.28	22.60	5.06	7.21	11.29
6	SoumikDas	11.14	2.68	10.18	10.60	8.65
7	Rishi	12.44	10.08	12.64	16.79	12.98
8	Agro_Geek	13.42	18.85	11.30	28.45	18.01
9	CropIQ	10.80	16.54	21.70	28.60	19.41
-	Baseline	5.86	5.71	10.62	8.80	7.74

Rank	Team	Okra	Radish	Mustard	Wheat	Average
1	ETTI 2 - VIS	1.06	0.83	2.70	3.38	1.99
2	AIgriTech	1.65	1.79	4.40	6.71	3.63
3	DeepLeaf	0.99	0.89	7.60	5.25	3.69
4	Resense	1.44	4.81	2.16	6.23	3.66
5	SoumikDas	3.06	1.33	3.33	7.20	3.73
6	PlantPixels	2.76	0.66	8.30	5.68	4.35
7	CropIQ	2.52	6.48	5.65	8.70	5.83
8	Rishi	18.50	6.89	4.15	5.18	8.68
9	Agro_Geek	1.85	3.98	7.38	30.41	10.90
-	Baseline	2.04	4.04	4.99	10.80	5.52

Tasks posed in the challenge

Plant Age Prediction using Multi-view Images

Participants must develop a model that predicts the age of a plant in days using multiple views of the same plant. The dataset for each crop should be used separately for training and validation. The dataset consists of images captured at five different height levels, and participants must incorporate all five levels in their model. The participants can vary the number of images per level to cover a 360° view. The accuracy of predictions will be assessed using MAE, with results reported separately for each crop. The final evaluation for this task will be based on the average MAE across all crops.

Leaf Count Estimation using Multi-view Images

Participants must build a model that counts the number of leaves on a plant using multiple views of the same plant. The dataset for each crop should be used separately for training and validation. The dataset consists of images captured at five different height levels, and participants must incorporate all five levels in their model. The participants can vary the number of images per level to cover a 360° view. The leaf count estimation will be assessed using MAE, with results reported separately for each crop. The final evaluation for this task will be based on the average MAE across all crops.

Challenge Overview

The challenge consists of two tasks: Plant Age Prediction and Leaf Count Estimation.
The dataset is provided in the Dataset section and includes separate zipped folders for four crops: Radish, Okra, Mustard, and Wheat, along with corresponding ground truth CSV files.
Both tasks are compulsory. Participants must train models separately for each crop and report results for each.
Each task requires four separate results, one per crop.
The performance metric for both tasks is Mean Absolute Error (MAE).
Participants can use any number of the 24 available multi-view images per plant at each level for training.
Separate models must be developed for each crop per task.
Leaderboard ranking is based on the average MAE across all four crops for each task, with separate leaderboards.
Each participant is allowed only five submissions.

Important Dates

Website and Call for Participation: 10th March 2025
Dataset Release: 15th March 2025
Registration Deadline: 30th April 2025
Testing Phase Submission Starts: 30th April 2025
Solution Submission: 13th June 2025
Evaluation Results Announcement: 12th July 2025
Grand Challenge Solution Submission: 30th July 2025
Notification: 24th July 2025
Camera-Ready Submission: 26th August 2025
Conference: 27th October - 31st October 2025

Submission Guidelines

Participants can use any number of multi-view images for training their models. Each participant is allowed up to five submissions. The final evaluation will be based on MAE score.

Submission details will be mailed to you after registration at your registered Email ID.

Paper Link GitHub

Advanced crop Monitoring

Revolutionizing Crop Monitoring with the GroMo Challenge

Effective plant growth monitoring is crucial for precision agriculture, plant breeding, and yield estimation. The GroMo challenge enhances crop monitoring by focusing on plant age estimation and multi-view leaf counting, leveraging phenotypic traits like leaf count and growth patterns for accurate plant health assessment.

Innovative Approaches in the GroMo Challenge

The GroMo challenge enhances crop monitoring with a multi-view time-series dataset, capturing plants from 24 angles and five heights to overcome occlusion challenges. This dataset enables dynamic plant growth analysis, improving leaf counting, age prediction, and phenotyping through computer vision and deep learning.

Impact on the Future of Crop Monitoring

The GroMo challenge advances crop monitoring by enhancing plant age estimation and leaf counting. High-accuracy models will optimize irrigation, fertilization, and pest management, improving yield prediction and agricultural sustainability. Its success can drive next-gen plant phenotyping and data-driven farming innovations.

We developed a Multi-View Vision Transformer (MVVT) model to process multiple images of the same plant from different perspectives, enabling improved plant age estimation and leaf counting. The model begins by applying separate patch embedding layers to each input image, converting them into a sequence of smaller patches. These embeddings are then enhanced with positional encodings to retain spatial information. A series of transformer encoder layers process these embeddings using self-attention mechanisms, capturing complex relationships between different views. The outputs from all views are concatenated and pooled before passing through a multi-layer perceptron (MLP) head, which predicts plant age or leaf count. This architecture leverages the power of self-attention to integrate multi-view information effectively, improving estimation accuracy. We have taken four images view to cover the 360° view.
Paper Link Github

Dataset

Dataset Overview

The dataset contains multi-view images of four crops—wheat, mustard, radish, and okra—captured for plant growth analysis in a controlled environment to ensure consistency across daily observations. A rotator device was used to capture images from various angles and heights, providing a comprehensive view. Covering the entire growth cycle of each crop, the dataset documents their structural changes over time for detailed analysis.

Crop	Plants	Max Days	Levels	Angles
Wheat	4	118	5	0° - 360° (step 15°)
Mustard	4	50	5	0° - 360° (step 15°)
Radish	5	59	5	0° - 360° (step 15°)
Okra	2	86	5	0° - 360° (step 15°)

Organizing Committee

Shreya Bansal

IIT Ropar

Ruchi Bhatt

IIT Ropar

Amanpreet Chander

IIT Ropar

Rupinder Kaur

IIT Ropar

Malya Singh

IIT Ropar

Dr. Mohan Kankanhalli

National University of Singapore

Abdulmotaleb El Saddik

University of Ottawa

Mukesh Kumar Saini

IIT Ropar

GROMO 25

LEADERBOARD

Age Prediction – MAE Scores

Tasks posed in the challenge

Plant Age Prediction using Multi-view Images

Leaf Count Estimation using Multi-view Images

GROMO 25

Challenge Overview

Important Dates

Submission Guidelines

Advanced crop Monitoring

Revolutionizing Crop Monitoring with the GroMo Challenge

Innovative Approaches in the GroMo Challenge

Impact on the Future of Crop Monitoring

Baseline

Dataset

Dataset Overview

Organizing Committee

Shreya Bansal

Ruchi Bhatt

Amanpreet Chander

Rupinder Kaur

Malya Singh

Dr. Mohan Kankanhalli

Abdulmotaleb El Saddik

Mukesh Kumar Saini

Get In Touch

Address

Links

Publication

GROMO 25

LEADERBOARD

Age Prediction – MAE Scores

Tasks posed in the challenge

Plant Age Prediction using Multi-view Images

Leaf Count Estimation using Multi-view Images

GROMO 25

Challenge Overview

Important Dates

Submission Guidelines

Advanced crop Monitoring

Revolutionizing Crop Monitoring with the GroMo Challenge

Innovative Approaches in the GroMo Challenge

Impact on the Future of Crop Monitoring

Baseline

Dataset

Dataset Overview

Select a Crop Dataset

Organizing Committee

Shreya Bansal

Ruchi Bhatt

Amanpreet Chander

Rupinder Kaur

Malya Singh

Dr. Mohan Kankanhalli

Abdulmotaleb El Saddik

Mukesh Kumar Saini

Get In Touch

Address

Links

Publication