Visualization-for-Darknet-network-structure-

[2019/09/04] adding node [conv_lstm]

[2019/05/14] Modify the number of output channel in residual operation (ElewiseSUM)

For instance, layer -1 (128 * 32 * 32) and layer 66 (256 * 32 * 32) (ch * w * h)

residual (layer -1 and layer 66) = 128 * 32 * 32

[2019/03/20] adding node [crnn]

Visualizing the darknet network structure by .cfg file, and also provide the output featrue map size for each layer.

How to use

Example: python3 main_DarknetVisualization.py

"import .cfg path" and "Result save format" can be modified in the 'main_DarknetVisualization.py' directly.

Result save format: default: 'png'. However, 'pdf' is better choice.

Output Example

YOLOv2

YOLOv3

Requirements

python3, graphviz