使用TensorFlow Lite 部署自定义对象检测模型

使用TensorFlow Lite 部署自定义对象检测模型^[1]

2022.03.05

一.訓練自定義模型^[4]

步驟:

1.收集數據

打標籤—標註精靈

2.訓練模型

二.集成TFLite模型的步驟

三.代碼實現:(教程)^[1]

下載源代碼

odml-pathways-main.zip\odml-pathways-main\object-detection\codelab2\android

starter – 核心關鍵代碼沒寫

final – 完整可運行的

引用:

[1] 使用 TensorFlow Lite 构建和部署自定义对象检测模(Android)

[2] 視頻講解參考

[3] TensorFlow Hub

[4] Train a salad detector with TFLite Model Maker

[5] Build and deploy custom object detection model with TensorFlow Lite|Workshop https://wwwyoutubecom/watch?v=vLxn5mOuWAk
[6] EfficientDet:Towards Scalable and Efficient Object Detection
https://aigoogleblogcom/2020/04/efficientdet-towards-scalable-andhtml
[7] WhatObjectCategories/Labels Are In COCO Dataset?https//techamikelive.com/node718/what-object-categories-labels-are-in-coco-dataset/
[8] MicrosoftCOCO:CommonObiectsinContexthttps://arxivorg/pdf/14050312pdf
[9] 目标检测中的AP，mAPhttps//zhuanlanzhihucom/p/888968686用YOLOv5+DeepSORT，打造实时多目标跟踪模型 https://mp.weixinqqcom/s/diPOaolDEf78t4AeU-i5bw