Awards of VDVWC We will award prizes to the top three teams: 5,000 INR for the first place, 3,000 INR for the second place, and 2,000 INR for the third place.
The three teams will be invited to contribute to the competition summary paper, which will be included in the proceedings of ICDEC 2024. Additionally, the top five participating teams will receive certificates from the ICDEC 2024 committee.
Competition Outline

Over the years, vehicle detection has predominantly been conducted on videos captured in natural or various weather conditions. However, there is still room for improvement in vehicle detection in various weather and lighting conditions. Furthermore, various videos are being captured for the detection of the vehicles, and the features of the vehicles including shape, color, texture, or visual characteristics are being extracted based on Region Of Interest (ROI) and classify them into different vehicle types.

The AVD-Dataset comprises 3,200 images of vehicles captured under various weather conditions. On June 15th, 2024, we will release approximately 2,600 images for training and 200 images for validation. All images are annotated in the YOLO format, and these annotations will be provided for both training and validation purposes.

Participants are hereby instructed to submit their codes in the GitHub link with proper documentation of the codes.

Each team can have up to 4 members. If a team submits more than once, the organizers will consider the last submission as the final entry. The winner will be determined based on the team achieving the highest mean Average Precision (mAP) and F1-Score performance on the test dataset.

Each team is required to provide two deliverables: a) GitHub link of the proposed model. b) A concise one-page report outlining the methodology employed in the detection algorithm and highlighting the novelty of the method. **Please note that the proposed machine learning model needs to be trained solely on the training dataset of the AVD-dataset.

   Competition Objectives

Establish a Benchmark for Vehicle Detection Algorithms: A critical objective of this competition is to establish a benchmark for evaluating and comparing the effectiveness of machine learning and deep learning-based vehicle detection algorithms. We use the mean Average Precision (mAP) metric to facilitate this. By pursuing these objectives, the competition aims to drive significant advancements in object detection technology and methodology, particularly under various weather and lighting conditions, thereby contributing to the broader field of computer vision.

For queries and suggestions, contact us at: sourajit.cse.ju@gmail.com, asfakali.etce@gmail.com