search

OlafenwaMoses / Traffic-Net

A dataset of traffic, fire and accident images for training deep learning models.
https://DeepQuestAI.com
MIT License
156 stars 51 forks source link

Wrong results with video file! #3

Open omarrajab94 opened 4 years ago

omarrajab94 commented 4 years ago

I get wrong results when i test the model with video file, but not, if i give pictures to the model. can you please show me, how can i make prediction for video frames. results look like: Fire : 100.0 Sparse_Traffic : 0.0 Dense_Traffic : 0.0 Accident : 0.0 video file is: https://www.pexels.com/video/traffic-flow-in-the-highway-2103099/

my Code:

Bildschirmfoto 2020-01-12 um 20 41 53
cross-hello commented 3 years ago

That is right, with a wrong result is normal behavior. The actual life view may differ from input photos, which mean the perspective angle, car size may different from one by one. you need to train the model by your inputs.