From 1f3c31e06aae6ffaae993f546cc8f103f9017ea9 Mon Sep 17 00:00:00 2001 From: Christoph Date: Sun, 19 May 2024 16:19:12 +0200 Subject: [PATCH] update ltr Readme --- ltr/README.md | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/ltr/README.md b/ltr/README.md index fa7a172b..9cbcc2f9 100644 --- a/ltr/README.md +++ b/ltr/README.md @@ -50,7 +50,8 @@ The framework consists of the following submodules. ### TaMOs The following setting files can be used to train the TaMOs tracker. In addition to the typical tracking datasets used for -single object trackers we further include [TAO](https://taodataset.org/), [YoutubeVOS](https://youtube-vos.org) and [ImagenetVid](http://image-net.org/) training data. +single object trackers we further include [TAO](https://taodataset.org/), [YoutubeVOS](https://youtube-vos.org) and [ImagenetVid](http://image-net.org/) training data. +When training with TAO we use the [BURST](https://github.com/Ali2500/BURST-benchmark) annotations since they provide a higher annotation frame rate. We converted those annotations to our own format [TaoBurst.json](https://drive.google.com/file/d/1q-x9BKa9cHYohh7le5sa9vk6is0Q2ljj/view?usp=sharing). - [tamos.tamos_resnet50](train_settings/tamos/tamos_resnet50.py): The default setting use for training with ResNet50 backbone. - [tamos.tamos_swin_base](train_settings/tamos/tamos_swin_base.py): The default setting use for training with SwinBase backbone. If needed, the weights of the SwinBase backbone can be downloaded [here](https://github.com/SwinTransformer/storage/releases/download/v1.0.0/swin_base_patch4_window12_384_22k.pth).