Individual Tree Delineation Contest 2026: Imagery-based Solutions

1. Background

Canopies play an essential role in biophysical functions of forest environments. The presence and structure of canopies exert significant influences on temperature, vaper concentration, and radiation regime in forests. At an individual tree level, the position, the size, and the geometry of a canopy are important structural characteristics that decides its functions. Hence, individual tree delineation (ITD) through high-resolution remote sensing (RS) image that provides the structural characteristics of the detected individual trees and their canopies has been becoming one of the most important topics in forest RS research during past two decades. The ITD Contest aims to clarify the state-of-the-art of the tree crown instance segmentation from high-resolution earth observation images through a benchmarking study of different methodological approaches.

2. Contest Overview

Participants of the contest will use high-resolution earth observation images and corresponding annotation of individual tree crown masks provided by the organizer to develop their own models, thus, to automatically segment individual tree crowns from other high-resolution images. The ITC results provided by the participants will be evaluated and benchmarked by the organizer using standardized the procedures. The organizer will analyze the outcomes and report the findings to reveal the state of the art and to discover the challenges.

For the latest News of the contest, please check the News page.

3. Participation

Please send following information by email to itdcontest.mspace@gmail.com for the contest link to participate, including:

• Team's name and members list (maximum 5, and list in order);
• Detailed information of each member, i.e., real name, email, affiliation, and position;
• Indicate a member for communication;
• adding a note: application for the ITD Contest 2026.

4. Tasks

The tasks of the image-based ITD contest include:

• to carry out image-based ITD. The participants extract the individual tree crown (ITC) masks using their automated methods/models based on the high-resolution aerial images provided by the organizer.
• to document the applied methods/models. The participants describe their applied methods/models, including the descriptions and the figures for the pipeline of methods or the structure of model. If a method is not yet published, a description with sufficient details (e.g., in 1-2-page A4) is required to support the understanding of the organizer. If the method has been published, please provide the reference and to provide a short method description, e.g., in 1-page A4. All the descriptions will be used for the joint publication of the results. Additional clarifications may be required from the organizer during the evaluation process.
• to submit the final results, including the ITD results and the method descriptions. Results without a sufficient documentation will not be considered in this contest for the prizes.

There are three phases in this contest, i.e., development, evaluation, winners determination phase. The detailed tasks of each phase include:

• In development phase, a public image-based ITD dataset with training, validation, and testing set is provided. Participants should develop the image-based ITD model based on the training and validation set, using either machine-learning- or deep-learning-based methods. In addition, the model's performance will be evaluated through submitting the ITD results of testing set to this platform. The accuracy (i.e., AP50 and AP75) and AP50 ranking will be displayed on the leaderboard.
• In evaluation phase, the image of evaluation data is released when this phase begins. Participants have one-week time to delineate the trees of evaluation data using their developed model, and submit the ITD results and method description.
• In winners determination phase, the top six teams from the evaluation phase are required to submit the Docker file and ITD models for results inference. The organizers will execute the ITD models and verify their results. Minor discrepancies between the testing and re-executed results are acceptable, but significant differences are not. If a team in the original top-six list fails to fulfill these requirements, the next highest-ranking team that submits all required materials will replace it. The final winner list is released after this phase.

5. Dataset

5.1. Overview

This contest provided the image-based ITD dataset with high-resolution remote sensing images collected from varies study sites around the world representing diver climate and forest types. It is the largest dataset, so far, with the richest diversity of scenarios to improve and evaluate the generalization and transferability of ITD methods.

5.2. Data format

• Image Data: The images are in “.png” format, including bands Red, Green, and Blue.
• Labels: The individual tree crown mask are stored in “.json” files, which are organized based on the MS COCO Format. The coordinates of the individual tree crown mask are in the image coordinate system, and are provided in the annotation file.

5.3. Composition

There are two sets of data are provided to the participants in the development and evaluation phases, respectively.

• In development phase, a public dataset, which consists of training, validation, and testing set, is provided to the participants. The training and evaluation set include both image and labels. The participants should develop their image-based ITD method based on the training and validation set. The testing set includes only images, which are collected from both the same and different study sites in the training set to evaluate the transferability of the applied models. The participants could submit the segmentation results of testing set to this platform to the feedbacks of average precision and ranking
• In evaluation phase, the evaluation data will be released to estimate the performance of the applied models from participants. The images are collected from both the same and different study sites in the training set, as well.

5.4. Download

The public dataset for development phase is now openly accessible on the Zenodo website. The evaluation data is released when this evaluation phase starts.

6. Submission

The submitted ITD results should be stored in a “.json” format file following the MS COCO Results Format.

In development phase, a file named testing_img_id.json is released with the testing set to appoint each image with an id. The participants should generate inference ITD results based on the ids.

In evaluation phase, a file with image-id correlation is released with the images, as well.

6.1. Results submission

The results submission in three phases vary:

• Phase 1: Development

Participants could submit the ITD results of testing set to get the accuracy score and ranking for several times. Thus, the participants are enabled to improve their method constantly. Submission in this phase is not compulsory.

• Phase 2: Evaluation

The evaluation data is released when evaluation phase begins. Participants should submit the ITD results of evaluation data to this platform. In addition, the ITD method description should be submitted before the deadline, as well. The AP50 ranking and description manuscript are the key metrics to determine the list for winners determination phase. The method description is required in a Microsoft word format file with the real names and afflictions of the authors of the method. Make sure the descriptions are clear for the organizer to understand. The latest submission of the testing results and methods description before the deadline will be applied for evaluations.

• Phase 3: Winners determination

The top six teams from the evaluation phase are required to submit the Docker file and ITD models for results inference. 

• File:

Participants should upload a ZIP file with a JSON file which stored the prediction results of ITC masks. The ZIP file should be name as "ITC_XXX_mask.zip", and the XXX should be the team's name. The JSON file should be named as "results.json", or it will lead the failure of scoring.

6.2. Supplemantory materials

Participants can provide supplementary materials if they wish. The supplementary materials are optional and will not necessarily be included in the final report and final publication.

7. Methods Evaluation

The organizers will evaluate the algorithm by comparing the submitted results with the ground truths. Identical evaluation approaches will be applied to the results of all participants.

7.1. Strategy

• The accuracy of ITC instance detection and shape delineation.
• The generalization and transferability of the model for multiple scenarios.

7.2. Metrics

The predictions are considered as true positive if their overlap with ground truth satisfies certain IoU (Intersection of Union) threshold, e.g., 50% or 75%. The evaluation metrics include AP50 and AP75.