使用 Ultralytics YOLO26 进行排队管理
什么是排队管理?#
使用 Ultralytics YOLO26进行队列管理,是指监控人群或车辆队列,以缩短等待时间并提高效率。YOLO26 会检测并跟踪等候区内的每个人,并实时报告队列长度,从而提升零售店、银行、机场和医疗机构等场所的客户满意度和运营效率。
观看: 如何使用 Ultralytics YOLO26 构建排队管理系统 | 实时队列分析 🚀
排队管理的优势#
- 缩短等待时间:排队管理系统能够高效组织队列,尽可能缩短客户的等待时间。客户等待更少、体验产品或服务的时间更多,因此满意度也会提高。
- 提高效率:实施排队管理有助于企业更有效地分配资源。通过分析队列数据并优化人员部署,企业可以简化运营、降低成本并提高整体生产力。
- 实时洞察:由 YOLO26 驱动的排队管理系统可即时提供队列长度和等待时间数据,帮助管理人员迅速做出明智决策。
- 提升客户体验:通过减少长时间等待带来的挫败感,企业可以提高客户满意度和忠诚度。
实际应用#
| 物流 | 零售 |
|---|---|
![]() | ![]() |
| 使用 Ultralytics YOLO26 管理机场售票柜台排队 | 使用 Ultralytics YOLO26 监控人群排队情况 |
使用 YOLO26 管理队列#
QueueManager 解决方案会逐帧统计指定区域内的跟踪对象,提供实时队列长度供你写入输出视频。传入覆盖等候区的多边形区域,然后通过 Python API 或 CLI 在视频源上运行该方案。
# Run a queue example
yolo solutions queue show=True
# Pass a source video
yolo solutions queue source="path/to/video.mp4"
# Pass queue coordinates
yolo solutions queue region="[(20, 400), (1080, 400), (1080, 360), (20, 360)]"QueueManager 参数#
下面是 QueueManager 参数表:
| 参数 | 类型 | 默认值 | 说明 |
|---|---|---|---|
model | str | None | Ultralytics YOLO 模型文件的路径。 |
region | list 或 dict | None | 定义感兴趣区域的点,可以是 (x, y) 元组列表,也可以是将区域名称映射到点列表的字典(仅支持多个 RegionCounter 区域)。如果为 None,则需要区域的解决方案会退回到预定义的默认区域。 |
line_width | int | 2 | 解决方案绘制的框、关键点和计数的线条粗细。 |
verbose | bool | True | 启用逐帧日志,记录输入形状、类别计数和处理速度。跟踪调用本身始终不会输出日志。 |
QueueManager 解决方案还支持 track 参数:
| 参数 | 类型 | 默认值 | 说明 |
|---|---|---|---|
tracker | str | 'tracktrack.yaml' | 指定要使用的跟踪算法。内置选项:botsort.yaml、bytetrack.yaml、ocsort.yaml、deepocsort.yaml、fasttrack.yaml、tracktrack.yaml。 |
conf | float | 0.25 | 设置检测的置信度阈值;较低的值允许跟踪更多物体,但可能会包含误检。 |
iou | float | 0.5 | 设置用于筛除重叠检测结果的交并比 (IoU) 阈值。 |
classes | list | None | 按类别索引筛选结果。例如,classes=[0, 2, 3] 只跟踪指定类别。 |
device | str | None | 指定推理设备(例如 cpu、cuda:0 或 0)。用户可以选择 CPU、特定 GPU 或其他计算设备来运行模型。 |
imgsz | int or tuple | 640 | 目标推理图像尺寸:正方形图像使用整数,或使用 (height, width) 对。图像尺寸会调整为模型步幅的整数倍。 |
max_det | int | 300 | 跟踪前每帧保留的最大检测数量,用于限制解决方案在密集场景中处理的物体数量。 |
quantize | int or str | None | 推理精度:对于非 CPU 上运行的 PyTorch 和 TorchScript 模型,使用 16 (FP16) 或 32/unset (FP32) 进行选择;其他格式使用其模型文件和运行时所选的精度。此参数替代已弃用的 half 参数。 |
此外,还提供以下可视化参数:
| 参数 | 类型 | 默认值 | 说明 |
|---|---|---|---|
show | bool | False | 如果 True,则在窗口中显示带标注的图像或视频。适用于在开发或测试期间即时查看结果。 |
show_conf | bool | True | 在每个检测结果的标签旁显示置信度分数,帮助了解模型对该检测结果的确信程度。 |
show_labels | bool | True | 在可视化输出中显示每个检测结果的标签,便于立即了解检测到的对象。 |
实施策略#
使用 YOLO26 实施排队管理时,请考虑以下最佳实践:
- 合理布置摄像头:调整摄像头位置,确保完整拍摄队列区域且不受遮挡。
- 定义合适的队列区域:根据场地的实际布局,仔细设定队列边界。
- 调整检测置信度:根据光照条件和人群密度微调置信度阈值。
- 与现有系统集成:将排队管理解决方案连接到数字标牌或员工通知系统,以实现自动响应。
常见问题#
要使用 Ultralytics YOLO26 实现实时排队管理,可以按以下步骤操作:
- 使用
cv2.VideoCapture捕获视频流。 - 为排队管理定义感兴趣区域 (ROI)。
- 使用 YOLO26 模型(例如
yolo26n.pt)和该区域初始化solutions.QueueManager。 - 处理视频帧以检测对象并管理队列。
下面是一个精简示例:
import cv2 from ultralytics import solutions cap = cv2.VideoCapture("path/to/video.mp4") queue_region = [(20, 400), (1080, 400), (1080, 360), (20, 360)] queuemanager = solutions.QueueManager( model="yolo26n.pt", region=queue_region, line_width=3, show=True, ) while cap.isOpened(): success, im0 = cap.read() if not success: break results = queuemanager(im0) cap.release() cv2.destroyAllWindows()你可以在 Ultralytics Platform上训练并部署队列管理解决方案所用的模型。
- 使用
使用 Ultralytics YOLO26 进行排队管理有多项优势:
- 缩短等待时间:高效组织队列,缩短客户等待时间并提升满意度。
- 提高效率:分析队列数据,优化人员部署和运营,从而降低成本。
- 实时洞察:逐帧提供实时队列人数统计,队列变长时便于及时介入。
- 可扩展性:可轻松扩展到零售店、机场和医疗机构等不同环境。
与基于 TensorFlow的检测器和 Detectron2 相比,Ultralytics YOLO26 在队列管理方面有多项优势:
- 实时性能:YOLO26 专为实时检测而设计,可以快速处理帧。
- 易用性:Ultralytics 覆盖从训练到部署的整个工作流程,可通过 Python 或 Ultralytics Platform 使用。
- 预训练模型:可使用多种预训练模型,缩短设置所需时间。
- 社区支持:丰富的文档和活跃的社区支持,让问题解决更加轻松。
了解如何开始使用 Ultralytics YOLO。
可以,Ultralytics YOLO26 能够管理各种类型的队列,包括机场和零售环境中的队列。通过为 QueueManager 配置特定区域和设置,YOLO26 可以适应不同的队列布局和密度。
机场场景示例:
queue_region_airport = [(50, 600), (1200, 600), (1200, 550), (50, 550)] queue_airport = solutions.QueueManager( model="yolo26n.pt", region=queue_region_airport, line_width=3, )Ultralytics YOLO26 用于各种现实世界的排队管理场景:
- 零售:监控结账队列,以缩短等待时间并提升客户满意度。
- 机场:管理售票柜台和安检处的队列,让乘客出行更加顺畅。
- 医疗保健:优化诊所和医院的患者流动。
- 银行:通过高效管理银行内的队列来提升客户服务。
查看我们的现实场景中的排队管理博客,了解计算机视觉如何改变各行业的队列监控。
评论18
提出问题、分享结果,或帮助其他人。
CharanPrasadK
Like in_counts and out_counts methods in objectCounter class, I want get Queue Count in real time to print and use on out side but the method you provided 'counts' in QueueManager class is refreshing to 0, so I made some changes in code in my local system used private variable to get the count out side in real time. How can I make it better?
Glenn JocherUltralytics(已编辑)@TechWolf21 to get the real-time queue count, read it from the results returned for each frame (
queue_count) instead of modifyingQueueManager:results = queue(im0) print(results.queue_count)For details, please refer to the Queue Management documentation.
Alexts166
Have a nice day. I made several queue areas, but I can't find the code to get a separate counter to count the total number of people who passed through these areas.
Glenn JocherUltralytics(已编辑)@Alexts166 hello!
Thank you for your question.
QueueManagercounts the people currently inside its region on each frame (results.queue_countorqueue.counts), so with several queue areas you can create oneQueueManagerper region and sum their counts:import cv2 from ultralytics import solutions cap = cv2.VideoCapture("path/to/video/file.mp4") assert cap.isOpened(), "Error reading video file" w, h, fps = (int(cap.get(x)) for x in (cv2.CAP_PROP_FRAME_WIDTH, cv2.CAP_PROP_FRAME_HEIGHT, cv2.CAP_PROP_FPS)) video_writer = cv2.VideoWriter("queue_management.avi", cv2.VideoWriter_fourcc(*"mp4v"), fps, (w, h)) # Define multiple queue regions queue_regions = [ [(20, 400), (1080, 404), (1080, 360), (20, 360)], [(30, 300), (1070, 304), (1070, 260), (30, 260)], ] # Initialize a QueueManager for each region (person class only) queues = [solutions.QueueManager(model="yolo26n.pt", region=region, classes=[0], show=False) for region in queue_regions] while cap.isOpened(): success, im0 = cap.read() if not success: print("Video frame is empty or video processing has been successfully completed.") break for queue in queues: queue(im0) # draws on im0 total_in_queues = sum(queue.counts for queue in queues) # people currently inside all queue regions print(f"People currently in the queue areas: {total_in_queues}") video_writer.write(im0) cap.release() video_writer.release() cv2.destroyAllWindows()Note this is a per-frame count of people currently in the regions, not a running total of everyone who has passed through. For cumulative in/out crossing counts, use
solutions.ObjectCounterinstead.For more details on queue management and additional examples, you can refer to our Queue Management Guide.
If you encounter any issues or have further questions, please provide a reproducible example, and ensure you are using the latest version of the packages. This will help us assist you more effectively.
Alexts166
Thanks for the code. I tweaked it a little to fit my needs. There is no summation of objects in queues. Maybe where is the error?
import cv2 from ultralytics import YOLO, solutions # Load the YOLOv8 model model = YOLO("best.pt") cap = cv2.VideoCapture("gf.mp4") assert cap.isOpened(), "Error reading video file" w, h, fps = (int(cap.get(x)) for x in (cv2.CAP_PROP_FRAME_WIDTH, cv2.CAP_PROP_FRAME_HEIGHT, cv2.CAP_PROP_FPS)) video_writer = cv2.VideoWriter("queue_management.avi", cv2.VideoWriter_fourcc(*"mp4v"), fps, (w, h)) # Define multiple queue regions queue_regions = [ [(20, 200), (1200, 200), (1200, 150), (20, 150)], [(20, 600), (1200, 600), (1200, 550), (20, 550)] ] # Initialize QueueManager for each region queues = [solutions.QueueManager( classes_names=model.names, reg_pts=region, line_thickness=3, fontsize=1.0, region_color=(255, 144, 31) ) for region in queue_regions] total_count = 0 while cap.isOpened(): success, im0 = cap.read() if success: tracks = model.track(im0, show=False, persist=True, verbose=False) for queue in queues: out = queue.process_queue(im0, tracks) total_count += queue.counts # Accumulate the count from each queue region color_black = (0, 0, 0) cv2.rectangle(im0, (15, 25), (250, 110), (0, 255, 255), -15) cv2.putText(im0, ('Summ - ') + str(total_count), (20, 50), cv2.FONT_HERSHEY_DUPLEX, 0.8, color_black, 2, cv2.LINE_AA) video_writer.write(im0) if cv2.waitKey(1) & 0xFF == ord("q"): break continue print("Video frame is empty or video processing has been successfully completed.") break cap.release() cv2.destroyAllWindows()
Glenn JocherUltralyticsHi! It looks like the issue might be with how the counts are being accumulated. Ensure that
queue.countsis correctly accessed and reset after each frame. Also, make sure you're using the latest version of the packages. If the problem persists, please provide more details. Happy coding!
QuifaAbas
Despite multiple attempts, I'm still facing the same problem. The code isn't generating the expected results. Even with the provided inputs, the results aren't found. Additionally, when trying to generate a video file, the output turns out to be empty. I input a 2.3 MB video file, and it is not an empty file. The code snippet I've been using is,
Install required packages
!pip install opencv-python torch
Import necessary libraries
from ultralytics import YOLO from ultralytics.solutions import distance_calculation import cv2
Load YOLO model
model = YOLO("/content/drive/MyDrive/YOLOv8_1/Distance/yolov8s.pt") # Update the path to yolov8s.pt names = model.model.names
Load video file
video_path = "/content/drive/MyDrive/YOLOv8_1/Distance/vid2.mp4" # Update the path to your video file cap = cv2.VideoCapture(video_path) assert cap.isOpened(), "Error reading video file" w, h, fps = (int(cap.get(x)) for x in (cv2.CAP_PROP_FRAME_WIDTH, cv2.CAP_PROP_FRAME_HEIGHT, cv2.CAP_PROP_FPS))
Video writer
output_video_path = "/content/drive/MyDrive/YOLOv8_1/Distance/distance_calculation.mp4" video_writer = cv2.VideoWriter(output_video_path, cv2.VideoWriter_fourcc(*'MP4V'), # Or try 'MP4V' for .mp4 files fps, (w, h))
Init distance-calculation obj
dist_obj = distance_calculation.DistanceCalculation() dist_obj.set_args(names=names, view_img=False) # Change view_img to False
while cap.isOpened(): success, im0 = cap.read() if not success: print("End of video or read error encountered.") break
tracks = model.track(im0, persist=True, show=False) # Check if any tracks were found if len(tracks) > 0: # Using len() to get the count of elements im0 = dist_obj.start_process(im0, tracks) # Debugging: Save frame only if it's not empty if im0 is not None and im0.any(): cv2.imwrite('/content/drive/MyDrive/YOLOv8_1/Distance/Copy of Copy of DJI_0083.JPG', im0) # Save a frame as an image for debugging video_writer.write(im0) else: print("Empty frame detected. Skipping saving to file.") else: print("No tracks found. Skipping frame processing.")Release resources
video_writer.release() cap.release()
Provide the download link for the output video
output_video_path
The specific error message I'm receiving is "Empty frame detected. Skipping saving to file. End of video or read error encountered.". Strangely, the resulting video file is only 258 bytes in size, indicating an issue with its generation.
In my pursuit to find a solution, I came across your YouTube video [YouTube Video Link: "https://www.youtube.com/watch?v=LE8am1QoVn4&ab_channel=Ultralytics", where the results showcased are exactly what I'm aiming for. Specifically, I'm keen on measuring the distance between two bounded boxes in both photos and videos.
Could you please provide a YOLOv8 Google Colab code for both video files and image files that addresses these issues and ensures error-free execution?Specifically, I would need two separate codes: one for processing video files and another for processing image files. Could you include instructions on what file paths I should replace in each code?
I've made several attempts to correct and run the code, but unfortunately, I keep encountering the same issue. It seems to persist despite my efforts. I hope you understand my actual problem and can provide a solution that resolves the issue. Could you please provide the complete code along with a detailed explanation to help me better understand and address the problem?
[ I want to attach your comment to my screenshot. How can I do that?I would like to include your comment in my screenshot for reference. Could you please provide guidance on how to attach your comment to the screenshot? ]
Glenn JocherUltralytics(已编辑)It looks like you're encountering a common issue with video processing where no tracks are found and your output frames are empty. Let's try a couple of things to resolve this:
-
Check Video Input: Confirm the video path and contents to ensure it's being read correctly. You have already done this, but double-check if the path is accessible and the video is not corrupted.
-
Model Loading: Verify the model path for loading the YOLO model. Ensure that
yolov8s.ptis the correct model file and path. -
Debugging Output: Try increasing the verbosity of your video processing to get more details:
tracks = model.track(im0, persist=True, show=True, verbose=True) -
Update Your Installation: Ensure your
ultralyticspackage is up to date as fixes are frequently released. Use:pip install --upgrade ultralytics -
Modify Export Code: If the issue persists, simplify the frame processing code to check if any frames are being written to the file at all:
while cap.isOpened(): ret, frame = cap.read() if not ret: break video_writer.write(frame) # write raw frames to see if writing is the issue -
Permissions and Path: Ensure that the output path for your video file is correct and your environment has write permissions for that directory.
Try these steps and see if the issue persists. If it does, you might want to check out a Colab example: Ultralytics YOLO Example Colab.
Regarding attaching a comment to a screenshot, if you're using screenshot tools like Snipping Tool or Snagit, you can usually add text directly on the image. Alternatively, you can use image editing software like Photoshop or free tools like GIMP to add text annotations.
Please let me know if these suggestions help resolve your issues or if there’s anything else I can assist you with!
-
QuifaAbas
I've attempted to run the code multiple times, but I'm consistently encountering errors. I ran the code that you provided with the suggested changes, but unfortunately, the issue persists. The code I've been using is,
Install required packages
!pip install opencv-python torch !pip install --upgrade ultralytics
Import necessary libraries
from ultralytics import YOLO from ultralytics.solutions import distance_calculation import cv2
Load YOLO model
model = YOLO("/content/drive/MyDrive/YOLOv8_1/Distance/yolov8s.pt") # Update the path to yolov8s.pt names = model.model.names
Load video file
video_path = "/content/drive/MyDrive/YOLOv8_1/Distance/vid2.mp4" # Update the path to your video file cap = cv2.VideoCapture(video_path) assert cap.isOpened(), "Error reading video file" w, h, fps = (int(cap.get(x)) for x in (cv2.CAP_PROP_FRAME_WIDTH, cv2.CAP_PROP_FRAME_HEIGHT, cv2.CAP_PROP_FPS))
Video writer
output_video_path = "/content/drive/MyDrive/YOLOv8_1/Distance/distance_calculation.mp4" video_writer = cv2.VideoWriter(output_video_path, cv2.VideoWriter_fourcc(*'MP4V'), # Or try 'MP4V' for .mp4 files fps, (w, h))
Init distance-calculation obj
dist_obj = distance_calculation.DistanceCalculation() dist_obj.set_args(names=names, view_img=False) # Change view_img to False
while cap.isOpened(): ret, frame = cap.read() if not ret: break video_writer.write(frame) # write raw frames to see if writing is the issue
tracks = model.track(im0, persist=True, show=True, verbose=True) # Check if any tracks were found if len(tracks) > 0: # Using len() to get the count of elements im0 = dist_obj.start_process(im0, tracks) # Debugging: Save frame only if it's not empty if im0 is not None and im0.any(): cv2.imwrite('/content/drive/MyDrive/YOLOv8_1/Distance/Copy of Copy of DJI_0083.JPG', im0) # Save a frame as an image for debugging video_writer.write(im0) else: print("Empty frame detected. Skipping saving to file.") else: print("No tracks found. Skipping frame processing.")Release resources
video_writer.release() cap.release()
Provide the download link for the output video
output_video_path
and the resulting output is,
Requirement already satisfied: torchvision>=0.9.0 in /usr/local/lib/python3.10/dist-packages (from ultralytics) (0.17.1+cu121) Requirement already satisfied: tqdm>=4.64.0 in /usr/local/lib/python3.10/dist-packages (from ultralytics) (4.66.2) Requirement already satisfied: psutil in /usr/local/lib/python3.10/dist-packages (from ultralytics) (5.9.5) Requirement already satisfied: py-cpuinfo in /usr/local/lib/python3.10/dist-packages (from ultralytics) (9.0.0) Requirement already satisfied: thop>=0.1.1 in /usr/local/lib/python3.10/dist-packages (from ultralytics) (0.1.1.post2209072238) Requirement already satisfied: pandas>=1.1.4 in /usr/local/lib/python3.10/dist-packages (from ultralytics) (2.0.3) Requirement already satisfied: seaborn>=0.11.0 in /usr/local/lib/python3.10/dist-packages (from ultralytics) (0.13.1) Requirement already satisfied: contourpy>=1.0.1 in /usr/local/lib/python3.10/dist-packages (from matplotlib>=3.3.0->ultralytics) (1.2.1) Requirement already satisfied: cycler>=0.10 in /usr/local/lib/python3.10/dist-packages (from matplotlib>=3.3.0->ultralytics) (0.12.1) Requirement already satisfied: fonttools>=4.22.0 in /usr/local/lib/python3.10/dist-packages (from matplotlib>=3.3.0->ultralytics) (4.51.0) Requirement already satisfied: kiwisolver>=1.0.1 in /usr/local/lib/python3.10/dist-packages (from matplotlib>=3.3.0->ultralytics) (1.4.5) Requirement already satisfied: numpy>=1.20 in /usr/local/lib/python3.10/dist-packages (from matplotlib>=3.3.0->ultralytics) (1.25.2) Requirement already satisfied: packaging>=20.0 in /usr/local/lib/python3.10/dist-packages (from matplotlib>=3.3.0->ultralytics) (24.0) Requirement already satisfied: pyparsing>=2.3.1 in /usr/local/lib/python3.10/dist-packages (from matplotlib>=3.3.0->ultralytics) (3.1.2) Requirement already satisfied: python-dateutil>=2.7 in /usr/local/lib/python3.10/dist-packages (from matplotlib>=3.3.0->ultralytics) (2.8.2) Requirement already satisfied: pytz>=2020.1 in /usr/local/lib/python3.10/dist-packages (from pandas>=1.1.4->ultralytics) (2023.4) Requirement already satisfied: tzdata>=2022.1 in /usr/local/lib/python3.10/dist-packages (from pandas>=1.1.4->ultralytics) (2024.1) Requirement already satisfied: charset-normalizer<4,>=2 in /usr/local/lib/python3.10/dist-packages (from requests>=2.23.0->ultralytics) (3.3.2) Requirement already satisfied: idna<4,>=2.5 in /usr/local/lib/python3.10/dist-packages (from requests>=2.23.0->ultralytics) (3.7) Requirement already satisfied: urllib3<3,>=1.21.1 in /usr/local/lib/python3.10/dist-packages (from requests>=2.23.0->ultralytics) (2.0.7) Requirement already satisfied: certifi>=2017.4.17 in /usr/local/lib/python3.10/dist-packages (from requests>=2.23.0->ultralytics) (2024.2.2) Requirement already satisfied: filelock in /usr/local/lib/python3.10/dist-packages (from torch>=1.8.0->ultralytics) (3.14.0) Requirement already satisfied: typing-extensions>=4.8.0 in /usr/local/lib/python3.10/dist-packages (from torch>=1.8.0->ultralytics) (4.11.0) Requirement already satisfied: sympy in /usr/local/lib/python3.10/dist-packages (from torch>=1.8.0->ultralytics) (1.12) Requirement already satisfied: networkx in /usr/local/lib/python3.10/dist-packages (from torch>=1.8.0->ultralytics) (3.3) Requirement already satisfied: jinja2 in /usr/local/lib/python3.10/dist-packages (from torch>=1.8.0->ultralytics) (3.1.3) Requirement already satisfied: fsspec in /usr/local/lib/python3.10/dist-packages (from torch>=1.8.0->ultralytics) (2023.6.0) Requirement already satisfied: nvidia-cuda-nvrtc-cu12==12.1.105 in /usr/local/lib/python3.10/dist-packages (from torch>=1.8.0->ultralytics) (12.1.105) Requirement already satisfied: nvidia-cuda-runtime-cu12==12.1.105 in /usr/local/lib/python3.10/dist-packages (from torch>=1.8.0->ultralytics) (12.1.105) Requirement already satisfied: nvidia-cuda-cupti-cu12==12.1.105 in /usr/local/lib/python3.10/dist-packages (from torch>=1.8.0->ultralytics) (12.1.105) Requirement already satisfied: nvidia-cudnn-cu12==8.9.2.26 in /usr/local/lib/python3.10/dist-packages (from torch>=1.8.0->ultralytics) (8.9.2.26) Requirement already satisfied: nvidia-cublas-cu12==12.1.3.1 in /usr/local/lib/python3.10/dist-packages (from torch>=1.8.0->ultralytics) (12.1.3.1) Requirement already satisfied: nvidia-cufft-cu12==11.0.2.54 in /usr/local/lib/python3.10/dist-packages (from torch>=1.8.0->ultralytics) (11.0.2.54) Requirement already satisfied: nvidia-curand-cu12==10.3.2.106 in /usr/local/lib/python3.10/dist-packages (from torch>=1.8.0->ultralytics) (10.3.2.106) Requirement already satisfied: nvidia-cusolver-cu12==11.4.5.107 in /usr/local/lib/python3.10/dist-packages (from torch>=1.8.0->ultralytics) (11.4.5.107) Requirement already satisfied: nvidia-cusparse-cu12==12.1.0.106 in /usr/local/lib/python3.10/dist-packages (from torch>=1.8.0->ultralytics) (12.1.0.106) Requirement already satisfied: nvidia-nccl-cu12==2.19.3 in /usr/local/lib/python3.10/dist-packages (from torch>=1.8.0->ultralytics) (2.19.3) Requirement already satisfied: nvidia-nvtx-cu12==12.1.105 in /usr/local/lib/python3.10/dist-packages (from torch>=1.8.0->ultralytics) (12.1.105) Requirement already satisfied: triton==2.2.0 in /usr/local/lib/python3.10/dist-packages (from torch>=1.8.0->ultralytics) (2.2.0) Requirement already satisfied: nvidia-nvjitlink-cu12 in /usr/local/lib/python3.10/dist-packages (from nvidia-cusolver-cu12==11.4.5.107->torch>=1.8.0->ultralytics) (12.4.127) Requirement already satisfied: six>=1.5 in /usr/local/lib/python3.10/dist-packages (from python-dateutil>=2.7->matplotlib>=3.3.0->ultralytics) (1.16.0) Requirement already satisfied: MarkupSafe>=2.0 in /usr/local/lib/python3.10/dist-packages (from jinja2->torch>=1.8.0->ultralytics) (2.1.5) Requirement already satisfied: mpmath>=0.19 in /usr/local/lib/python3.10/dist-packages (from sympy->torch>=1.8.0->ultralytics) (1.3.0) WARNING ⚠️ Environment does not support cv2.imshow() or PIL Image.show()
NameError Traceback (most recent call last) in <cell line: 31>() 35 video_writer.write(frame) # write raw frames to see if writing is the issue 36 ---> 37 tracks = model.track(im0, persist=True, show=True, verbose=True) 38 39 # Check if any tracks were found
NameError: name 'im0' is not defined
Unfortunately, I'm unable to attach a screenshot directly. Could you please provide the complete code to resolve this issue?
I did try my older code, which is,
Install required packages
!pip install opencv-python torch
Import necessary libraries
from ultralytics import YOLO from ultralytics.solutions import distance_calculation import cv2
Load YOLO model
model = YOLO("/content/drive/MyDrive/YOLOv8_1/Distance/yolov8s.pt") # Update the path to yolov8s.pt names = model.model.names
Load video file
video_path = "/content/drive/MyDrive/YOLOv8_1/Distance/vid2.mp4" # Update the path to your video file cap = cv2.VideoCapture(video_path) assert cap.isOpened(), "Error reading video file" w, h, fps = (int(cap.get(x)) for x in (cv2.CAP_PROP_FRAME_WIDTH, cv2.CAP_PROP_FRAME_HEIGHT, cv2.CAP_PROP_FPS))
Video writer
output_video_path = "/content/drive/MyDrive/YOLOv8_1/Distance/distance_calculation.mp4" video_writer = cv2.VideoWriter(output_video_path, cv2.VideoWriter_fourcc(*'MP4V'), # Or try 'MP4V' for .mp4 files fps, (w, h))
Init distance-calculation obj
dist_obj = distance_calculation.DistanceCalculation() dist_obj.set_args(names=names, view_img=False) # Change view_img to False
while cap.isOpened(): success, im0 = cap.read() if not success: print("End of video or read error encountered.") break
tracks = model.track(im0, persist=True, show=False) # Check if any tracks were found if len(tracks) > 0: # Using len() to get the count of elements im0 = dist_obj.start_process(im0, tracks) # Debugging: Save frame only if it's not empty if im0 is not None and im0.any(): cv2.imwrite('/content/drive/MyDrive/YOLOv8_1/Distance/Copy of Copy of DJI_0083.JPG', im0) # Save a frame as an image for debugging video_writer.write(im0) else: print("Empty frame detected. Skipping saving to file.") else: print("No tracks found. Skipping frame processing.")Release resources
video_writer.release() cap.release()
Provide the download link for the output video
output_video_path
and the resulting output was ,
Requirement already satisfied: nvidia-nccl-cu12==2.19.3 in /usr/local/lib/python3.10/dist-packages (from torch) (2.19.3) Requirement already satisfied: nvidia-nvtx-cu12==12.1.105 in /usr/local/lib/python3.10/dist-packages (from torch) (12.1.105) Requirement already satisfied: triton==2.2.0 in /usr/local/lib/python3.10/dist-packages (from torch) (2.2.0) Requirement already satisfied: nvidia-nvjitlink-cu12 in /usr/local/lib/python3.10/dist-packages (from nvidia-cusolver-cu12==11.4.5.107->torch) (12.4.127) Requirement already satisfied: MarkupSafe>=2.0 in /usr/local/lib/python3.10/dist-packages (from jinja2->torch) (2.1.5) Requirement already satisfied: mpmath>=0.19 in /usr/local/lib/python3.10/dist-packages (from sympy->torch) (1.3.0) WARNING ⚠️ Environment does not support cv2.imshow() or PIL Image.show()
requirements: Ultralytics requirement ['lapx>=0.5.2'] not found, attempting AutoUpdate... Collecting lapx>=0.5.2 Downloading lapx-0.5.8-cp310-cp310-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl (1.7 MB) ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 1.7/1.7 MB 14.6 MB/s eta 0:00:00 Requirement already satisfied: Cython>=0.29.32 in /usr/local/lib/python3.10/dist-packages (from lapx>=0.5.2) (3.0.10) Requirement already satisfied: numpy>=1.21.6 in /usr/local/lib/python3.10/dist-packages (from lapx>=0.5.2) (1.25.2) Installing collected packages: lapx Successfully installed lapx-0.5.8
requirements: AutoUpdate success ✅ 8.7s, installed 1 package: ['lapx>=0.5.2'] requirements: ⚠️ Restart runtime or rerun command for updates to take effect
0: 384x640 (no detections), 566.5ms Speed: 29.6ms preprocess, 566.5ms inference, 4702.4ms postprocess per image at shape (1, 3, 384, 640) Empty frame detected. Skipping saving to file.
0: 384x640 (no detections), 436.2ms Speed: 3.1ms preprocess, 436.2ms inference, 0.9ms postprocess per image at shape (1, 3, 384, 640) Empty frame detected. Skipping saving to file.
0: 384x640 (no detections), 410.8ms Speed: 3.4ms preprocess, 410.8ms inference, 1.5ms postprocess per image at shape (1, 3, 384, 640) Empty frame detected. Skipping saving to file.
0: 384x640 1 potted plant, 412.6ms Speed: 3.1ms preprocess, 412.6ms inference, 15.0ms postprocess per image at shape (1, 3, 384, 640) Empty frame detected. Skipping saving to file.
0: 384x640 1 potted plant, 428.9ms Speed: 3.2ms preprocess, 428.9ms inference, 1.3ms postprocess per image at shape (1, 3, 384, 640) Empty frame detected. Skipping saving to file.
0: 384x640 1 potted plant, 545.6ms Speed: 3.8ms preprocess, 545.6ms inference, 1.9ms postprocess per image at shape (1, 3, 384, 640) Empty frame detected. Skipping saving to file.
0: 384x640 (no detections), 546.6ms Speed: 3.0ms preprocess, 546.6ms inference, 1.3ms postprocess per image at shape (1, 3, 384, 640) Empty frame detected. Skipping saving to file.
0: 384x640 (no detections), 626.9ms Speed: 3.6ms preprocess, 626.9ms inference, 1.2ms postprocess per image at shape (1, 3, 384, 640) Empty frame detected. Skipping saving to file.
0: 384x640 (no detections), 631.8ms Speed: 3.4ms preprocess, 631.8ms inference, 1.2ms postprocess per image at shape (1, 3, 384, 640) Empty frame detected. Skipping saving to file. End of video or read error encountered. /content/drive/MyDrive/YOLOv8_1/Distance/distance_calculation.mp4
The specific error message I'm receiving is "Empty frame detected. Skipping saving to file. End of video or read error encountered.". Strangely, the resulting video file is only 258 bytes in size, indicating an issue with its generation.
Glenn JocherUltralytics(已编辑)Hi there! 👋 It seems like you're encountering a persistent issue with detecting objects in your video file. Let's try a couple of adjustments:
-
Ensure Correct Model Load: Double-check if the path to your model (yolov8s.pt) is correct and the model is appropriately loaded.
-
Verify Video Content: Sometimes, if the video does not include the objects the model is trained to detect, it may not find any tracks. Confirm that the target objects are present in the video.
-
Frame Handling: Right before calling the solution (
distancecalculator(frame)), ensure thatframeis your actual frame variable fromret, frame = cap.read(). -
Debugging Output: Increase logging for each step and verify each component's output, such as the captured frames and the model's response.
Here's a simplified snippet to recheck each part:
from ultralytics import solutions distancecalculator = solutions.DistanceCalculation(model="yolo26n.pt", show=True) # tracking is handled internally while cap.isOpened(): ret, frame = cap.read() if not ret: print("Failed to read frame.") break # Debugging: Show frame dimension print(f"Frame dimensions: {frame.shape}") results = distancecalculator(frame) video_writer.write(results.plot_im) # Cleanup cap.release() video_writer.release() print("Released all resources.")Adjust the debugging prints as necessary to check the outputs directly. Let me know how it goes, and if there's anything else I can help with! 🚀
-
QuifaAbas(已编辑)
I have zero knowledge in YOLOv8, as I'm an undergraduate student in Electrical Engineering. My final year project topic is 'Unmanned Aerial Systems (UAS) for Enhanced Vegetation Management in Electric Distribution Networks'. Initially, I aimed to detect vegetation and power lines in single frames. I successfully achieved this using YOLOv8 in Google Colab for images. However, I encountered difficulties when applying the same approach to videos, particularly in detecting power lines accurately.
To enhance the accuracy of power line detection, I plan to collect more photos using a drone. Additionally, I aim to measure the distance between vegetation and power lines. Despite my efforts, I've faced repeated challenges in implementing these tasks.
I have shared a link to my Google Drive , which contains my files. Within this folder, I've included one of the best detection images, the detected video, the final output, and my code.
I kindly request two separate complete codes—one for image tracking and another for video tracking. As I have minimal knowledge about where to make changes in the code, it would greatly assist me if you could run the code using my data and provide the necessary adjustments.
Glenn JocherUltralytics(已编辑)Hello! 🚀 It's great to see you using YOLOv8 for your project on unmanned aerial systems for vegetation management. Since you're encountering issues with power line detection in videos, here are a few suggestions:
-
Data Collection: Collecting more diverse images of power lines and vegetation at different times and conditions will definitely help in improving the model's accuracy.
-
Annotation Quality: Ensure that your additional images are accurately annotated to improve learning during training.
-
Model Training for Video: Although you've successfully used YOLOv8 for single frame detections, video processing can be trickier due to motion and different angles. Consider using techniques like temporal smoothing or frame-to-frame tracking to enhance accuracy.
Here's a basic outline on how you could set up your code for video processing using your own data:
import cv2 from ultralytics import YOLO # Load your pre-trained model model = YOLO("/path/to/your/model.pt") # Setup the video capture cap = cv2.VideoCapture("/path/to/your/video.mp4") while cap.isOpened(): ret, frame = cap.read() if not ret: break # Detect objects in the frame results = model.predict(frame) # Optionally, visualize the results annotated_frame = results[0].plot() # frame with the detections drawn cv2.imshow('YOLO Detections', annotated_frame) if cv2.waitKey(1) & 0xFF == ord('q'): break cap.release() cv2.destroyAllWindows()For measuring distances between detected objects like vegetation and power lines, you'd have to post-process the detection bounding boxes from YOLOv8. Calculating distances could be based on known scales or additional sensors like LiDAR data if your drone setup permits.
If you have any more questions or need further assistance with your code, feel free to ask. Best of luck with your exciting project! 🌿⚡
-
QuifaAbas(已编辑)
I've tried multiple times, but I haven't been successful in getting the distance calculation code to work properly.
-
Could you please ensure that the distance calculation code is implemented correctly for both my image and video files? [I have asked multiple times for you in providing separate code for processing my image and video files. However, you have only provided code for the video file. As I have zero knowledge in YOLOv8 coding, it's challenging for me to adapt the provided code for my image files. I kindly request your understanding of my situation and would greatly appreciate separate code for both image and video files.]
-
Once you've done so, Could you please provide the results for both my video and image files, which I have attached in the Google Drive folder?
-
Can you share these results with me via your Google Drive link?
I have a tight deadline to complete my project within this week, and your assistance would be immensely valuable. Thank you for understanding my situation. I am seeking results that accurately measure the distance between the vegetation bounded box and powerline bounded box in both my input image and input video. The desired outcome is to achieve results that closely resemble the accuracy and presentation demonstrated in your YouTube video (https://www.youtube.com/watch?v=LE8am1QoVn4&t=5s&ab_channel=Ultralytics).
I have clearly articulated my problem and outlined my specific requests. I kindly request your attention to my three requests and urge you to help me resolve my problem. 🥰
-
Glenn JocherUltralytics(已编辑)Hello! Thanks for sharing the details and your Google Drive link. It sounds like you're facing a challenge with adapting the distance calculation code specifically for images, along with your existing video processing.
For processing images separately, you can follow a similar approach as with videos but adapted for single-frame use. Here is a simplified code snippet that processes individual images:
from ultralytics import YOLO import cv2 # Load the model model = YOLO('path/to/your/model.pt') # Load an image image_path = 'path/to/your/image.jpg' image = cv2.imread(image_path) # Predict using the model results = model.predict(image) # Display detected objects with bounding boxes results[0].show()For the videos, if the provided code doesn't meet expectations, it helps to first manually check a few frames to understand what's missing or incorrect. This could be due to various factors such as the model's sensitivity, training data, or the quality/resolution of the video frame.
I've noticed your tight deadline and understand the importance of precise and timely responses. Please check the code above, and let's ensure it aligns with your project's requirements. If you need adjustments or further assistance, feel free to let me know! 🌟
QuifaAbas(已编辑)
Once again, I am encountering a problem. I have attached a screenshot of the error I'm facing, along with the image for which I want to measure the distance between the vegetation and powerline. Initially, could you please provide me with the tracked image? Subsequently, I will require the results image corresponding to the attached image. Following that, kindly provide the code necessary to achieve this task.
Glenn JocherUltralytics(已编辑)@QuifaAbas hello! Thanks for reaching out and sharing the details and your screenshot. It seems like you're looking to measure distances between detected objects in an image, which is a great application of YOLOv8.
For measuring distances, we would typically process the output of the detection model to obtain bounding boxes, and then calculate the distances based on these coordinates.
Here's a simplified approach to get you started with the image processing:
from ultralytics import YOLO import cv2 # Load your pre-trained model model = YOLO('path/to/your/model.pt') # Load your image image = cv2.imread('path/to/your/image.jpg') # Perform detection results = model.predict(image) # Extract bounding boxes (for example, for vegetation and power lines) result = results[0] for box in result.boxes: name = result.names[int(box.cls)] if name == 'vegetation': # replace with your actual class veg_box = box.xyxy[0].tolist() # x1, y1, x2, y2 elif name == 'power line': # replace with your actual class pline_box = box.xyxy[0].tolist() # x1, y1, x2, y2 # Calculate distances between objects (simplified Euclidean distance) distance = ((veg_box[0] - pline_box[0])**2 + (veg_box[1] - pline_box[1])**2)**0.5 # Draw results on image annotated_image = result.plot() # Save or display the annotated image cv2.imwrite('output.jpg', annotated_image) cv2.imshow('Distance Detection', annotated_image) cv2.waitKey(0) cv2.destroyAllWindows()For video files, you can follow a similar approach by processing each frame for detections and computing distances as outlined above.
If you need further customization or assistance, we're here to help! 🚀
QuifaAbas(已编辑)
Once again, I am encountering a problem. I have attached a screenshot of the error I'm facing, along with the image for which I want to measure the distance between the vegetation and powerline. My objective is to detect and measure distances between power lines and vegetation. Initially, could you please provide me with the tracked image? Subsequently, I will require the results image corresponding to the attached image. Following that, kindly provide the code necessary to achieve this task.
Glenn JocherUltralytics(已编辑)Hello! 🚀 It looks like you've run into some challenges with error handling and distance measurements between detected objects in your images. Let's tackle this step by step.
To help you measure the distances between detected vegetation and power lines in the image, you can utilize the coordinates of the bounding boxes provided by the YOLOv8 detections. Here's a simple code snippet to get you started:
from ultralytics import YOLO import cv2 # Load your model model = YOLO('path/to/your/yolov8-model.pt') # Load your image image = cv2.imread('path/to/your/image.jpg') # Detect objects results = model.predict(image) # Draw boxes and labels on the image results[0].show() # Save or display the image cv2.imwrite('tracked_image.jpg', results[0].plot()) # Extract bounding boxes for specific classes veg_boxes = [b.xyxy[0].tolist() for b in results[0].boxes if results[0].names[int(b.cls)] == 'Vegetation'] pline_boxes = [b.xyxy[0].tolist() for b in results[0].boxes if results[0].names[int(b.cls)] == 'Power Line'] # Assuming you want the distance between the closest points, calculate Euclidean distance from scipy.spatial import distance if veg_boxes and pline_boxes: veg_point = veg_boxes[0][:2] # Example: top-left corner of the first vegetation box pline_point = pline_boxes[0][:2] # Example: top-left corner of the first powerline box dist = distance.euclidean(veg_point, pline_point) print(f"Distance between closest points: {dist} pixels")Make sure to modify the
path/to/your/yolov8-model.ptandpath/to/your/image.jpgaccording to your files. This code will print the Euclidean distance between the top left corners of the first detected vegetation and power line boxes.I hope this helps! Start with this basic functionality and you can tune further based on your project needs. Let me know if there are specific aspects you want to adjust or if you encounter further issues! 🌿⚡

