add readme
Browse files- README.md +26 -1
- extract.py +23 -0
README.md
CHANGED
|
@@ -1 +1,26 @@
|
|
| 1 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
# OVMOT Detections
|
| 2 |
+
|
| 3 |
+
OVMOT Detections provides detailed results on open-vocabulary multi-object tracking benchmarks. The dataset is tailored for researchers and practitioners focusing on novel object detection and tracking challenges.
|
| 4 |
+
|
| 5 |
+
## Dataset Overview
|
| 6 |
+
- **Name:** OVMOT Detections
|
| 7 |
+
- **Task:** Open-Vocabulary Multi-Object Tracking
|
| 8 |
+
- **Data Type:** Detection results
|
| 9 |
+
- **Processing Note:** All detection results are post processed with **NMS 50**
|
| 10 |
+
- **Detector:** [GLEE Plus](https://github.com/FoundationVision/GLEE)
|
| 11 |
+
- **Benchmark:** [BFT](https://github.com/George-Zhuang/NetTrack), [OVT-B](https://github.com/Coo1Sea/OVT-B-Dataset), and [OVTAO](https://github.com/siyuanliii/TETA)
|
| 12 |
+
|
| 13 |
+
## Intended Use
|
| 14 |
+
This dataset is designed for:
|
| 15 |
+
- Evaluating and benchmarking multi-object tracking algorithms.
|
| 16 |
+
- Research on open-vocabulary object tracking.
|
| 17 |
+
- Development of new tracking methods in real-world scenarios.
|
| 18 |
+
|
| 19 |
+
## Dataset Structure
|
| 20 |
+
- **Files:** The dataset includes detection results alongside potential metadata and annotations.
|
| 21 |
+
- **Others:** Please refer to https://github.com/George-Zhuang/LBM for dataset structure.
|
| 22 |
+
|
| 23 |
+
## Limitations
|
| 24 |
+
- The use of a non-max suppression threshold of 50 may impact detection performance.
|
| 25 |
+
|
| 26 |
+
|
extract.py
ADDED
|
@@ -0,0 +1,23 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import os
|
| 2 |
+
import tarfile
|
| 3 |
+
|
| 4 |
+
|
| 5 |
+
folders = ['bft', 'ovt-b', 'ovtao']
|
| 6 |
+
base_dir = os.path.abspath(os.path.dirname(__file__))
|
| 7 |
+
|
| 8 |
+
for folder in folders:
|
| 9 |
+
folder_path = os.path.join(base_dir, folder)
|
| 10 |
+
if not os.path.isdir(folder_path):
|
| 11 |
+
print(f"{folder_path} does not exist.")
|
| 12 |
+
continue
|
| 13 |
+
|
| 14 |
+
for file_name in os.listdir(folder_path):
|
| 15 |
+
if file_name.endswith('.tar.gz'):
|
| 16 |
+
file_path = os.path.join(folder_path, file_name)
|
| 17 |
+
try:
|
| 18 |
+
with tarfile.open(file_path, 'r:gz') as tar:
|
| 19 |
+
tar.extractall(path=folder_path)
|
| 20 |
+
os.remove(file_path)
|
| 21 |
+
print(f"{file_path} extracted and removed.")
|
| 22 |
+
except Exception as e:
|
| 23 |
+
print(f"Process {file_path} error: {e}")
|