Object Detection using neural networks

0 1
  • 0 Collaborators

detecting the object in real time(say from a video) and thus making the output video with the name of an object around it ...learn more

Artificial Intelligence

Groups
DeepLearning

Overview / Usage

We make an architecture of convolution neural network and we train it with some open source dataset to make a real-time output like from videos by dividing it in frames and then process it using SSD i.e. single shot multi-box detection, in which we divide our image in a rectangular grid. It is a better alternative to YOLO algorithm.
we have also use deconvolution to deal with the scale problem.
In order to avoid training by own the alternative "ssd300_mAP_77.43_v2.pth" can be used as it is a free and open source. or we can go for GPU.

Comments (1)