search

John1liu / YOLOV5-DeepSORT-Vehicle-Tracking-Master

In this project, urban traffic videos are collected from the middle section of Xi 'an South Second Ring Road with a large traffic flow, and interval frames are extracted from the videos to produce data sets for training and verification of YOLO V5 neural network. Combined with the detection results, the open-source vehicle depth model data set is used to train the vehicle depth feature weight file, and the deep-sort algorithm is used to complete the target tracking, which can realize real-time and relatively accurate multi-target recognition and tracking of moving vehicles.
86 stars 11 forks source link

how to add yolov5/ssd features together #2

Open besbesmany opened 3 years ago

besbesmany commented 3 years ago

I need to add any other deeplearning technique like ssd , vgg16 after yolov5

how to take features from yolov5 and merge them with other deeplearning technique

can you help me please or send me link or code for this?

John1liu commented 3 years ago

I think VGG16 maybe too small to assign the feature from yolov5. What is your goal of merging these networks?

besbesmany commented 3 years ago

I want to enhance yolov5 prediction result , I am searching for new idea to apply