Replies: 1 comment
-
The issue you're encountering is likely related to the model's training and the diversity of your dataset. YOLOv5 is designed for real-time object detection in both images and videos, but there are some nuances to consider when applying it to video data. The reason your model performs well on the test set is probably because the frame closely resembles the images used during training. But, videos introduce temporal aspects that static images lack. The model might struggle when the position or orientation of a car in a particular video frame doesn't align with what it learned from the training data. It is possible to attain a good accuracy by training YOLOv5 on images too, as long as your dataset is diverse enough. |
Beta Was this translation helpful? Give feedback.
-
I have trained a
car
detection model with yolov5 by training onimages
withcar
and itsbboxes
.It performs well on the test set.
But when I test this model on a
video
, this trained yolov5 model always missed some car frames.So shouldn't we use yolov5 for video?
What is the difference between object detection for image and video with yolov5?
Should I train yolov5 by training on videos?
Beta Was this translation helpful? Give feedback.
All reactions