vod_based_multi_object_tracking
https://github.com/nimeshs54/vod_based_multi_object_tracking
Science Score: 26.0%
This score indicates how likely this project is to be science-related based on various indicators:
-
○CITATION.cff file
-
✓codemeta.json file
Found codemeta.json file -
✓.zenodo.json file
Found .zenodo.json file -
○DOI references
-
○Academic publication links
-
○Academic email domains
-
○Institutional organization owner
-
○JOSS paper metadata
-
○Scientific vocabulary similarity
Low similarity (6.6%) to scientific vocabulary
Repository
Basic Info
- Host: GitHub
- Owner: Nimeshs54
- License: apache-2.0
- Language: Python
- Default Branch: main
- Size: 107 MB
Statistics
- Stars: 0
- Watchers: 1
- Forks: 0
- Open Issues: 0
- Releases: 0
Metadata Files
README.md
Video Object Detection-Based Tracking
Introduction
Multi-Object Tracking (MOT) is a fundamental computer vision task involving the detection and tracking of objects in video sequences. Tracktor, an advanced MOT algorithm, leverages a Faster R-CNN object detector. However, tracking multiple objects in video sequences poses challenges, including:
- Degradation of object appearances due to rapid motion.
- Camera defocus.
- Pose variations.
In this repository, we propose a novel approach to enhance multi-object tracking performance by integrating Sequence Level Semantics Aggregation (SELSA)a Video Object Detection (VID) techniqueinto the Tracktor framework. SELSA replaces Faster R-CNN in our pipeline to address these challenges, improving tracking accuracy and robustness. Our method, tested on the MOTChallenge dataset, achieves:
- A 3.6% enhancement in Multiple Object Tracking Accuracy (MOTA).
- A 1.2% increase in Higher Order Tracking Accuracy (HOTA) compared to the conventional Tracktor with Faster R-CNN.
Installation
Follow the installation guide provided by MMTracking Installation.
Data Preparation
- Download the MOTChallenge dataset from MOTChallenge.net.
- Refer to the Data Setup Documentation for proper dataset preparation.
Results
Performance Comparison on the MOT15 Validation Dataset
| Method | Precision | Recall | IDF1 | MOTA | HOTA | | -------------------- | --------- | ------ | ---- | ---- | ---- | | Tracking with FRCNN | 85.2 | 81.6 | 68.3 | 66.6 | 52.9 | | Tracking with SELSA | 86.7 | 85.0 | 64.9 | 70.5 | 53.2 |
This repository showcases the potential of combining SELSA with the Tracktor framework to push the boundaries of multi-object tracking performance.
Owner
- Name: Nimesh Singh
- Login: Nimeshs54
- Kind: user
- Repositories: 1
- Profile: https://github.com/Nimeshs54
GitHub Events
Total
- Push event: 1
Last Year
- Push event: 1
Dependencies
- actions/checkout v2 composite
- actions/setup-python v2 composite
- codecov/codecov-action v1.0.10 composite
- codecov/codecov-action v2 composite
- actions/checkout v2 composite
- actions/setup-python v2 composite
- actions/checkout v2 composite
- actions/setup-python v2 composite
- actions/checkout v2 composite
- actions/setup-python v2 composite
- pytorch/pytorch ${PYTORCH}-cuda${CUDA}-cudnn${CUDNN}-devel build
- cython *
- numpy *
- myst_parser *
- sphinx ==4.0.2
- sphinx-copybutton *
- sphinx_markdown_tables *
- mmcls >=0.16.0,<1.0.0
- mmcv-full >=1.6.1,<1.7.0
- mmdet >=2.19.1,<3.0.0
- mmcls *
- mmcv *
- mmdet *
- torch *
- torchvision *
- attributee *
- dotty_dict *
- einops *
- lap *
- matplotlib *
- mmcls >=0.16.0,<1.0.0
- motmetrics *
- packaging *
- pandas <=1.3.5
- pycocotools *
- scipy <=1.7.3
- seaborn *
- terminaltables *
- tqdm *
- asynctest * test
- codecov * test
- flake8 * test
- interrogate * test
- isort ==4.3.21 test
- kwarray * test
- pytest * test
- ubelt * test
- xdoctest >=0.10.0 test
- yapf * test