Pothole Detection Challenge

Overview

Potholes present a significant risk to road safety and vehicle durability, demanding prompt detection to ensure effective infrastructure upkeep. This AI competition invites participants to design advanced computer vision models capable of automatically identifying potholes from road imagery.

The objective is to develop precise and scalable solutions that can empower intelligent systems, aiding municipalities, navigation platforms, and autonomous vehicles in enhancing road safety and optimizing maintenance workflows.

---

Practice Skills

In this challenge, you will gain hands-on experience with:

• Python.
• Deep Learning.
• Object detection.

---

Evaluation

Goal

Your objective is to detect potholes in the test images.

Metric

The Mean Average Precision (mAP) (at IoU=0.5) will be used for both the public and private test sets.

---

Submission Format

Submission guide

Submit a csv file with two columns:

• id: The unique identifier (image name) for each image from test set (e.g., 0001.jpg).
• predict_label: A string representing the predicted results in the image, this string has format: class confidence_score x_center y_center box_width box_height:
  • class: object category (1 : pothole).
  • confidence_score: the confidence score of the prediction (range 0-1).
  • x_center, y_center: normalized coordinates of the bounding box center.
  • box_width, box_height: normalized width and height of the bounding box.

Note: x_center and box_width are normalized by image_width, y_center and box_height are normalized by image_height.

Example:

• An object

ID, predict_label
0001.jpg, 1 0.927 0.191 0.808 0.208 0.103

• Two objects

ID, predict_label
0002.jpg, 1 0.927 0.191 0.808 0.208 0.103 1 0.555 0.123 0.138 0.338 0.443

The challenge includes two types of submissions:

Public Submission

• Use the public testing dataset (provided in the Data section) to predict your public submission results.

Private Submission

• Submit your trained model.
• It will be evaluated on a private testing dataset.
• Important: Your code must recursively scan the entire data directory (Code section for details).

---

Challenge Timeline

Challenge Phase	Duration	Start Date	End Date	Description
Registration	2 weeks	03/12/2025	16/12/2025	Participant registration period
Training Phase	2 weeks	17/12/2025	30/12/2025	50% dataset released for training & public submission
Validation Phase	2 weeks	31/12/2025	13/01/2026	Remaining 50% dataset released for training & public submission
Final Submission	1 week	14/01/2026	20/01/2026	Private submission phase for final model
Judging Phase	2 week	21/01/2026	03/02/2026
Result Announcement	1 day	04/02/2026	04/02/2026	Final results & prize winners announced

• Note: Timeline EST

---

Change log

🗓️ Public Dataset (for public evaluation)

• ⏳ Dataset Phase 1 (Training phase): 17/12/2025
• ⏳ Dataset Phase 2 (Validation phase): 31/12/2025

🗓️ Final Submission

• ⏳ Upload your private solution: 14/01/2026

---

Prizes

• Total Challenge Rewards: 4,500 AIOZ tokens.
• Special Reward: 2,000 AIOZ tokens.
• Top 5 Winners: 500 AIOZ tokens each.