邱棚
eb89f036bd
|
||
|---|---|---|
| .. | ||
| README.md | ||
| build_engine.py | ||
| build_engines.sh | ||
| graphsurgeon.patch-4.2 | ||
| graphsurgeon.patch-4.2.2 | ||
| graphsurgeon.patch-4.4 | ||
| install.sh | ||
| install_pycuda.sh | ||
| libflattenconcat.so.5 | ||
| libflattenconcat.so.6 | ||
| ssd_mobilenet_v1_coco.pb | ||
| ssd_mobilenet_v1_egohands.pb | ||
| ssd_mobilenet_v2_coco.pb | ||
| ssd_mobilenet_v2_egohands.pb | ||
README.md
Reference:
Sources of the trained models:
-
'ssd_mobilenet_v1_coco.pb' and 'ssd_mobilnet_v2_coco.pb': This is just the 'frozen_inference_graph.pb' file in ssd_mobilenet_v1_coco_2018_01_28.tar.gz and ssd_mobilenet_v2_coco_2018_03_29.tar.gz, i.e. 2 of the trained models in TensorFlow 1 Detection Model Zoo.
-
'ssd_mobilenet_v1_egohands.pb' and 'ssd_mobilenet_v2_egohands.pb': These models are trained using my Hand Detection Tutorial code. After training, just run the export.sh script to generated the frozen graph (pb) files.
-
I've also added support for ssd_inception_v2_coco in the code. You could download the .pb by following the link.