- Fine-tune models like ssd_mobilenet、faster_rcnn models in COCO dataset for detecting tongues
- All elementary codes are from TensorFlow Object Detection API
- The Chinese version tutorial of my little project is here https://mp.weixin.qq.com/s?__biz=MzIyMjE5Njk1Mw==&mid=2651247687&idx=1&sn=737d94e4503d1bc0585137aea49f922d&scene=19#wechat_redirect
- 'tongue_inference_graph_latest.pb' is the latest model I trained, so you can use it for tongue detection directly
- the folders 'faster_rcnn_inception_resnet_v2_atrous_coco_11_06_2017' and 'ssd_mobilenet_v1_coco_11_06_2017' are empty, you have to download the inference graph respectively into these folders
- The dataset is little, you can follow the link above to enrich your dataset by labelImge tool, then rebuild .csv and .record files
DevEvn Python-v3(3.6)
:
- tensorflow==1.9.0
Hardware Equipment:
- 4 core CPU
- 24gb memory
- 11gb video memory
- 484gb/s bandwidth of video memory
- 11.34tflop (single precision)
- 2T hard disk
Result: