thi23459112

Why cv2.imshow fuction is not supported by Cog?

1

If I include the line cv2.imshow("test", im0) in my code, why can't predict.py execute? If I delete that line, it runs successfully. The cv2.imshow("test", im0) line is simply displaying the...

我這邊說的自定義 是指不用detect.py的方式（without detect.py） 好比說可以用torch.hub加載或者還是未來可以規劃成v9 pip化 裡面的code就可以讓使用者自訂內容？ 謝謝

你好 想請問你的代碼，能否用yolov8加入它本身有提供bytetrack追蹤功能? 另外，區域計數能否跟越線那樣 數字只會越來越大(累計功能) 而非只數區域內的數量 謝謝

How to calculate TP, FN, FP, and TN?

1

I used the following command line to run YOLO detection validation: `yolo classify val model=path/to/best-cls.pt ` ``` confusion_matrix: curves: [] curves_results: [] fitness: 0.9385964870452881 keys: ['metrics/accuracy_top1', 'metrics/accuracy_top5'] results_dict: {'metrics/accuracy_top1': 0.8771929740905762,...

Is There Any Plan for YOLOv11?

6

I would like to ask the YOLOv8 team, YOLO has already reached the tenth generation, and I am wondering if your team has any plans to release the 11th generation....

Please Support the rf-detr Model

3

https://github.com/roboflow/rf-detr The latest and most powerful object detection model, rf-detr, has been officially released. Please integrate this model into the ultralytics.

Why Is Gaze-Detection Video Inference So Slow

1

I’m using a GPU with NVIDIA GeForce RTX 2060, but when I run the gaze-detection-video.py script for gaze detection, the inference process is extremely slow. I expected better performance from...

WARNING ⚠️ NMS time limit 2.050s exceeded

4

After running boxmot for about three hours, the terminal kept showing: > WARNING ⚠️ NMS time limit 2.050s exceeded > WARNING ⚠️ NMS time limit 2.050s exceeded > WARNING ⚠️...

AttributeError: 'NoneType' object has no attribute 'create_execution_context'

2

作者你好 目前我使用你的專案，將自己訓練yolo11的best.pt轉成engine，然後執行的時候，出現 classname 錯誤，她似乎是用coco本身的classes 如下圖所示 請問該如何解決呢，謝謝

新增 FastTracker 追蹤方法

2

麻煩考慮新增 FastTracker 追蹤法。 這是該方法的論文： 📄 https://arxiv.org/pdf/2508.14370 GitHub 專案： 💻 https://github.com/Hamidreza-Hashempoor/FastTracker?tab=readme-ov-file FastTracker 無需使用昂貴的 ReID 網路，而是透過幾何與場景感知（scene awareness），在複雜交通場景中實現精準且即時的追蹤，同時能有效降低 ID Switch 的發生率。 建議可以考慮將此方法納入比較或實驗項目。 謝謝！