Are you sure you want to delete this task? Once this task is deleted, it cannot be recovered.
You can not select more than 25 topics
Topics must start with a chinese character,a letter or number, can include dashes ('-') and can be up to 35 characters long.
jhvisionchen
fd08150c36
|
3 years ago | |
|---|---|---|
| cfg | 3 years ago | |
| data | 3 years ago | |
| examples | 3 years ago | |
| include | 3 years ago | |
| python | 3 years ago | |
| src | 3 years ago | |
| .gitignore | 3 years ago | |
| LICENSE | 3 years ago | |
| Makefile | 3 years ago | |
| README.md | 3 years ago | |
| camera_yolov3.sh | 3 years ago | |
| image_yolov1.sh | 3 years ago | |
| image_yolov2.sh | 3 years ago | |
| image_yolov3.sh | 3 years ago | |
| video_yolov3.sh | 3 years ago | |
README.md
yoloDet
1. Description
The object detection algorithm framework just for you only look once(YOLO).
1.1 yolo v1
The network layers [ based on Alexnet ] that be used in yolov1 include:
convolutional, batch_normalize, activation, maxpool, local, dropout, connected, detection.
The convolutional layer is used to extracting features.
The batchnorm layer is used to speed up convergence.
The activation layer is used to improve the ability of characterization.
The maxpool layer is used to space invariant deformation.
The local layer is used to improve feature diversity.
The dropout lyaer is used to improve model generalization ability
The connectd layer is used to classify or regression results.
The detection layer is used to calculate losses and diviations.
1.convolutional layer
2. batchnorm layer
3. activation layer
4. maxpool layer
5. local layer
6. detection layer
1.2 yolo v2
The network layers [ based on Darknet-19 ] that be used in yolov2 include:
convolutional, batch_normalize, activation, maxpool, route, reorg, region.
The route layer is used to concatenate the feature map.
The reorg layer is used to transform the feature map space into channel.
The region layer is used to get object bounding boxes.
1. route layer
2. reorg layer
3. region layer
1.3 yolo v3
The network layers [ based on Darknet-53 ] that be used in yolov3 include:
convolutional, batch_normalize, activation, shortcut, route, upsample, yolo.
The shortcut layer is used to add feature maps of the same resolution.
The upsample layer is used to enlarge the resolution of feature map.
The yolo layer is used to get object bounding boxes.
1. shortcut layer
2. upsample layer
3. yolo layer
2. Installation and uninstallation
- make -j 8
- make clean
3. Run and shut down
No Description
C INI Jupyter Notebook Cuda C++ other