Park Vision is a desktop app built by a WPI Interactive Qualifying Project to help the National Park Service understand visitor use in Acadia National Park, using still images and video already captured by trail cameras deployed in the park. Rather than manually reviewing thousands of photos and hours of footage, it gives you a toolkit of independent analyses you can mix and match as needed: detect and track people, bicycles, and vehicles; estimate direction of travel and speed; read license plates to compute dwell time; and compute per-image occupancy counts, exporting whichever results you've generated to CSV.
Read the full writeup in Report.pdf for methodology, evaluation, and results.
Grab the build for your platform from the latest release.
- Download
ParkVisionSetup.exeand run it. - Follow the installer; it adds a Start Menu shortcut and an uninstaller.
- Download
park-vision-macos-arm64.zipand unzip it. - Move
ParkVision.appwherever you'd like (e.g./Applications) and open it.
- Download
park-vision-linux-x86_64.zipand unzip it. - Run the extracted binary (
chmod +x ParkVisionfirst if needed):./ParkVision
Park Vision's GUI (src/ui) is built on top of a set of standalone
command-line tools (src/detection, src/processing, src/utility) that
can also batch-process video footage end-to-end outside of the GUI (YOLO
detection → entity tracking → speed/direction/dwell-time computation →
summary report).
- Python >= 3.10 (CI targets 3.13)
uvfor dependency management- Tesseract OCR on
PATH(used as a fallback to OCR the timestamp burned into camera footage, when a frame's filename/mtime doesn't already give a reliable one) - Git LFS: Model weights (
*.pt) and sample media are tracked via LFS (see.gitattributes); rungit lfs pullafter cloning if large files show up as pointer text
git clone https://github.com/ElliotScher/Acadia-AI-E26.git
cd Acadia-AI-E26
uv syncuv sync installs all dependencies declared in pyproject.toml, including
PyTorch, Ultralytics YOLO, OpenCV, PySide6, and SQLAlchemy.
Model weights live under models/:
| Path | Purpose |
|---|---|
models/generic/yolo26s.pt, yolov8s.pt |
General-purpose YOLO object detection (people, bikes, vehicles) |
models/generic/yolo26s-pose.pt |
Pose keypoints, used for pedestrian direction |
models/license_plate/license-plate.pt |
License plate localization |
models/vehicle_direction/last.pt |
Vehicle-pose model (fine-tuned on CarFusion) used for vehicle direction; see src/utility/fetch_vehicle_pose_weights.py if it needs to be re-downloaded |
uv run python src/ui/main.pyEach stage below is also runnable independently, and each script's own
--help documents its full option set:
src/detection/video_yolo.py/image_yolo.py: Run YOLO over raw video frames or images, producing a JSON detection reportsrc/processing/video_entityprofiler.py: Track unique entities across a video's frames from that report, export the best frame per entity, and compute relative/absolute speed and directionsrc/processing/report_recalibrator.py: Recompute absolute speed on an existing entity-profiler report against a different reference entity, without reprocessing videosrc/utility/report_summarizer.py: Turn an entity-profiler (or recalibrator) report into summary statistics: entity counts by type and direction, per-video breakdowns, speed statisticssrc/processing/video_plateextractor.py→plate_dwellprofiler.py: Detect and OCR license plates across video frames, then match repeated plate readings to compute vehicle dwell timesrc/processing/image_occupancyprofiler.py: Re-identify and track entities chronologically across a folder of images to compute occupancy counts
Utility scripts under src/utility/ also include benchmarks
(vehicle_speed_benchmark.py, vehicle_direction_benchmark.py,
pedestrian_direction_benchmark.py, plate_ocr_benchmark.py,
speed_distribution_comparator.py) that validate these algorithms against
labeled/ground-truth datasets, and scraper.py for bulk-downloading photos
from a Spypoint camera account.
uv run pytestTests live under tests/, mirroring the src/ package layout, and run
against sample media in tests/data/.
Desktop builds are produced with PyInstaller (main.spec), driven by
.github/workflows/build.yml for Linux, Windows, and macOS:
uv run pyinstaller main.specOn Windows, the resulting binary is wrapped into an installer with
Inno Setup via installer/main.iss.
Licensed under the GNU AGPLv3.
Acadia National Park final transportation plan / Environmental impact statement. (2019,
March). National Park Service.
https://parkplanning.nps.gov/document.cfm?parkID=203&documentID=94071
Ahad, A., Kidwai, F. A., & Alqadhi, S. (2026). AI-augmented real-time parking occupancy
detection and prediction: Integrating computer vision with user behavioral insights
for Delhi’s mixed-traffic ecosystem. Transportation.
https://doi.org/10.1007/s11116-026-10767-1
Albers, J. L., Wildhaber, M. L., Green, N. S., Struckhoff, M. A., & Hooper, M. J.
(2023). Visitor use and activities detected using trail cameras at forest
restoration sites. Ecological Restoration, 41(4), 199–212.
https://doi.org/10.3368/er.41.4.199
Arshad, J., Ijaz, C. A., Yousaf, A., Habib, S., Rehman, A. U., Abid, H., & Asif, R. M.
(2022). Implementation of an intelligent parking system. 2022 International
Conference on Engineering and Emerging Technologies (ICEET), 1–11.
https://doi.org/10.1109/ICEET56468.2022.10007136
Bai, L., Wu, C., Xie, F., & Wang, Y. (2021). Crowd density detection method based on
crowd gathering mode and multi-column convolutional neural network. Image and Vision
Computing, 105, 104084. https://doi.org/10.1016/j.imavis.2020.104084
Bai, Y., Zou, Q., Chen, X., Li, L., Ding, Z., & Chen, L. (2023). Extreme low-resolution
action recognition with confident spatial-temporal attention transfer. International
Journal of Computer Vision, 131(6), 1550–1565.
https://doi.org/10.1007/s11263-023-01771-4
Butoto, J., Liu, X., & Sando, T. (2026). Dwell time estimation using periodic image
captures and deep learning. The International FLAIRS Conference Proceedings, 39(1).
https://doi.org/10.32473/flairs.39.1.141550
Choi, W., Chao, Y.-W., Pantofaru, C., & Savarese, S. (2014). Discovering Groups of
People in Images. In D. Fleet, T. Pajdla, B. Schiele, & T. Tuytelaars (Eds.),
Computer Vision – ECCV 2014 (Vol. 8692, pp. 417–433). Springer International
Publishing. https://doi.org/10.1007/978-3-319-10593-2_28
Eco-Counter. (n.d.). Counting solutions to understand user flows. Retrieved April 30,
2026, from https://www.eco-counter.com/solutions/counting-solutions
Elek, O., Thomas, M. M., & Forbes, A. (2019). Learning patterns in sample distributions
for Monte Carlo variance reduction (arXiv:1906.00124; Version 1). arXiv.
https://doi.org/10.48550/arXiv.1906.00124
Fleming, D. (2016). As popular Acadia turns 100, there’s no room at the top. TCA
Regional News.
https://www.pressherald.com/2016/07/03/as-popular-acadia-turns-100-theres-no-room-at-the-top/
Ilesanmi, A. E., & Ilesanmi, T. O. (2021). Methods for image denoising using
convolutional neural network: A review. Complex & Intelligent Systems, 7(5),
2179–2198. https://doi.org/10.1007/s40747-021-00428-4
Jocher, G., Qiu, J., Liu, M., Lyu, S., Akyon, F. C., & Kalfaoglu, M. E. (2026).
Ultralytics YOLO26: Unified Real-Time End-to-End Vision Models (Version 1). arXiv.
https://doi.org/10.48550/ARXIV.2606.03748
Jones, T. E., Yang, Y., & Yamamoto, K. (2018). Comparing automated and manual visitor
monitoring methods: Integrating parallel datasets on Mount Fuji’s North Face.
Journal of Park and Recreation Administration, 36(1), 22–38.
https://doi.org/10.18666/JPRA-2018-V36-I1-7976
Li, N. (2023). Ethical considerations in artificial intelligence: A comprehensive
disccusion from the perspective of computer vision. SHS Web of Conferences, 179,
04024. https://doi.org/10.1051/shsconf/202317904024
Liu, Z., Dai, C., & Li, X. (2024). An electric bicycle tracking algorithm for improved
traffic management. Heliyon, 10(13), e32708.
https://doi.org/10.1016/j.heliyon.2024.e32708
Louridas, P., & Ebert, C. (2016). Machine learning. IEEE Software, 33(5), 110–115.
https://doi.org/10.1109/MS.2016.114
Lu, R., Zhang, H., & Zhang, Z. (2025). Revealing urban functionality through vehicle
dwell patterns: A data-driven approach. 2025 IEEE 28th International Conference on
Intelligent Transportation Systems (ITSC), 2544–2549.
https://doi.org/10.1109/ITSC60802.2025.11423278
Lune, H., & Berg, B. L. (2017). Qualitative research methods for the social sciences
(Ninth edition, global edition). Pearson.
Lupp, G., Kantelberg, V., Förster, B., Honert, C., Naumann, J., Markmann, T., & Pauleit,
S. (2021). Visitor counting and monitoring in forests using camera traps: A case
study from Bavaria (Southern Germany). Land (Basel), 10(7), 736.
https://doi.org/10.3390/land10070736
Managing congestion: A toolkit for parks. (2020, December). National Park Service.
https://www.nps.gov/orgs/1548/upload/Congestion_Management_2021-508.pdf
Manning, R., Jacobi, C., & Marion, J. L. (2006). Recreation monitoring at Acadia
National Park. The George Wright Forum, 23(2), 59–72. JSTOR.
Miguel, J., Mendonça, P., Quelhas, A., Caldeira, J. M. L. P., & Soares, V. N. G. J.
(2024). The development of a prototype solution for collecting information on
cycling and hiking trail users. Information, 15(7), 389.
https://doi.org/10.3390/info15070389
Mohammed, S., Budach, L., Feuerpfeil, M., Ihde, N., Nathansen, A., Noack, N., Patzlaff,
H., Naumann, F., & Harmouch, H. (2025). The effects of data quality on machine
learning performance on tabular data. Information Systems, 132, 102549.
https://doi.org/10.1016/j.is.2025.102549
Monika, Singh, P., & Chand, S. (2023). Computer vision-based framework for pedestrian
movement direction recognition. Journal of Intelligent & Fuzzy Systems, 44(5),
8015–8027. https://doi.org/10.3233/JIFS-223283
Monitoring guidebook. (2019, June). Interagency Visitor Use Management Council.
https://visitorusemanagement.nps.gov/Content/documents/508_final_Monitoring_Guidebook_Edition_One_IVUMC.pdf
Nadadur, D., Haralick, R. M., & Gustafson, D. E. (2005). A Bayesian framework for noise
covariance estimation using the facet model. IEEE Transactions on Image Processing,
14(11), 1902–1917. https://doi.org/10.1109/TIP.2005.854480
National Park Service. (2025). Annual park ranking report for recreation visits in:
2025. Integrated Resource Management Applications.
https://irma.nps.gov/Stats/SSRSReports/National%20Reports/Annual%20Park%20Ranking%20Report%20(1979%20-%20Last%20Calendar%20Year)
Quarterly acreage reports. (2026, March 31). National Park Service.
https://www.nps.gov/subjects/lwcf/acreagereports.htm
Razzaque, T., Hussain, H., Ahmmad, R., & Siddique, S. (2024). Enhancing Safety and
Collision Avoidance in Autonomous Vehicles through Pose Estimation Techniques. 2024
13th International Conference on Electrical and Computer Engineering (ICECE),
385–390. https://doi.org/10.1109/ICECE64886.2024.11024749
Singh, S., & Singh, K. (2025). Pattern recognition and image segmentation based on some
novel fuzzy similarity measures. Journal of Experimental & Theoretical Artificial
Intelligence, 37(8), 1453–1480. https://doi.org/10.1080/0952813X.2024.2440662
Tian, B., Tang, M., & Wang, F.-Y. (2015). Vehicle detection grammars with partial
occlusion handling for traffic surveillance. Transportation Research Part C:
Emerging Technologies, 56, 80–93. https://doi.org/10.1016/j.trc.2015.02.020
Viñals, M. J., Orozco Carpio, P. R., Teruel, P., & Gandía-Romero, J. M. (2024).
Real-time monitoring of visitor carrying capacity in crowded historic streets
through digital technologies. Urban Science, 8(4), 190.
https://doi.org/10.3390/urbansci8040190
Waelen, R. A. (2023). The ethics of computer vision: An overview in terms of power. AI
and Ethics, 4(2), 353–362. https://doi.org/10.1007/s43681-023-00272-x
Wang, Q., Liu, T., & Li, R. (2025). Artificial intelligence and environmental
sustainability: Investigating the AI‐EKC Nexus for SDG 7 and SDG 13.
https://doi.org/10.1002/sd.70294
Watson, A. E., Cole, D. N., Turner, D. L., & Reynolds, P. S. (2000). Wilderness
recreation use estimation: A handbook of methods and systems. United States
Department of Agriculture Forest Service.
https://www.fs.usda.gov/rm/pubs/rmrs_gtr056.pdf
Xue, W., Sun, J., Liang, F., Hou, J., Yang, Y., Shang, W., Chen, X., Gradoni, G., &
Huang, Y. (2025). Jensen–Shannon divergence hypothesis test for determining
reverberation chamber field distribution. IEEE Transactions on Antennas and
Propagation, 73(9), 6855–6870. https://doi.org/10.1109/TAP.2025.3574921
Yohannes, E., Lin, C.-Y., Shih, T. K., Thaipisutikul, T., Enkhbat, A., & Utaminingrum,
F. (2023). An improved speed estimation using deep homography transformation
regression network on monocular videos. IEEE Access, 11, 5955–5965.
https://doi.org/10.1109/ACCESS.2023.3236512
Zhao, G., Takafumi, M., Shoji, K., & Kenji, M. (2012). Video based estimation of
pedestrian walking direction for pedestrian protection system. Journal of
Electronics (China), 29(1–2), 72–81. https://doi.org/10.1007/s11767-012-0814-y
Zualkernan, I., Dhou, S., Judas, J., Sajun, A. R., Gomez, B. R., & Hussain, L. A.
(2022). An IoT system using deep learning to classify camera trap images on the
edge. Computers, 11(1), 13. https://doi.org/10.3390/computers11010013