Hi,
I am trying to implement YoloV3 for weapon detection in video surveillance system.
I have trained for Gun, Knife Classes and the system is detecting decently on good resolution images but not on low quality images from video surviellance footages.
My machine hardware specification used for YoloV3 Training:
RAM: 16 GB
Processor: i7-8750H CPU @2.20 GHz x 12
Graphics: GeForce GTX 1060
CUDA: 10.0.130
CUDA Cores: 1280
OS: Ubuntu 18.04.1 LTS
Hello @HidayathullaShaik, what was your average loss when you stopped the training? Also, I am curious about weapon image dataset. Can you share the resources for them?
Hi, I am trying to implement YoloV3 for weapon detection in video surveillance system. I have trained for Gun, Knife Classes and the system is detecting decently on good resolution images but not on low quality images from video surviellance footages.
My machine hardware specification used for YoloV3 Training: RAM: 16 GB Processor: i7-8750H CPU @2.20 GHz x 12 Graphics: GeForce GTX 1060 CUDA: 10.0.130 CUDA Cores: 1280 OS: Ubuntu 18.04.1 LTS
I have already implemented the suggestions at: https://github.com/AlexeyAB/darknet#how-to-improve-object-detection
Please let me know if there is any thing that I could do to improve the prediction of video surveillance systems.
Thanks, Hidayath