增量小样本目标检测
其他文章列表:
- Sylph模型单机多卡与单机单卡训练的区别。
- 配置论文《Sylph: A Hypernetwork Framework for Incremental Few-shot Object
Detection》代码过程中遇到的bug修复记录。 - 未完待续…
文章目录
-
增量小样本目标检测 - 其他文章列表:
- 一、背景:
- 二、定义:
- 三、相关论文:
- 01 Incremental Few-Shot Object Detection.**[CVPR 2020](https://dblp.org/db/conf/cvpr/cvpr2020.html#Perez-RuaZHX20): 13843-13852
- 02 Towards Generalized and Incremental Few-Shot Object Detection.** [CoRR abs/2109.11336](https://dblp.org/db/journals/corr/corr2109.html#abs-2109-11336) (2021)
- 03 Few-Shot Batch Incremental Road Object Detection via Detector Fusion.** [ICCVW 2021](https://dblp.org/db/conf/iccvw/iccvw2021.html#TambwekarAMS21): 3063-3070
- 04 Class-Incremental Few-Shot Object Detection.** [CoRR abs/2105.07637](https://dblp.org/db/journals/corr/corr2105.html#abs-2105-07637) (2021)
- 05 Sylph: A Hypernetwork Framework for Incremental Few-shot Object Detection.**[CVPR 2022](https://dblp.org/db/conf/cvpr/cvpr2022.html#YinPL22): 9025-9035
- 06 Incremental Few-Shot Object Detection for Robotics.** [ICRA 2022](https://dblp.org/db/conf/icra/icra2022.html#LiZTF0T0VL22): 8447-8453
- 07 Meta-Learning-Based Incremental Few-Shot Object Detection.** [IEEE Trans. Circuits Syst. Video Technol. 32(4)](https://dblp.org/db/journals/tcsv/tcsv32.html#ChengWL22): 2158-2169 (2022)
- 08 Incremental few-shot object detection via knowledge transfer.** [Pattern Recognit. Lett. 156](https://dblp.org/db/journals/prl/prl156.html#FengZYL22): 67-73 (2022)
- 09 Incremental Few-Shot Object Detection via Simple Fine-Tuning Approach.** [ICRA 2023](https://dblp.org/db/conf/icra/icra2023.html#ChoiK23): 9289-9295
- 10 Novel-Registrable Weights and Region-Level Contrastive Learning for Incremental Few-shot Object Detection.** [ICONIP (11) 2023](https://dblp.org/db/conf/iconip/iconip2023-11.html#TangMQHZQCJGL23): 400-412
- 11 Incremental-DETR: Incremental Few-Shot Object Detection via Self-Supervised Learning.** [AAAI 2023](https://dblp.org/db/conf/aaai/aaai2023.html#Dong0DL23): 543-551
- 12 Class Incremental Robotic Pick-and-Place via Incremental Few-Shot Object Detection.** [IEEE Robotics Autom. Lett. 8(9)](https://dblp.org/db/journals/ral/ral8.html#DengZHZW23): 5974-5981 (2023)
- 13 . Incremental Few-Shot Object Detection with scale- and centerness-aware weight generation. [Comput. Vis. Image Underst. 235](https://dblp.org/db/journals/cviu/cviu235.html#ZhangYQZL23): 103774 (2023)
- 14 Few-Shot Incremental Object Detection in Aerial Imagery via Dual-Frequency Prompt.** [IEEE Trans. Geosci. Remote. Sens. 62](https://dblp.org/db/journals/tgrs/tgrs62.html#LuDLZWSF24): 1-17 (2024)
- 15 An adaptive lightweight small object detection method for incremental few-shot scenarios of unmanned surface vehicles.** [Eng. Appl. Artif. Intell. 133](https://dblp.org/db/journals/eaai/eaai133.html#WangJLLCL24): 107989 (2024)
- 16 InfRS: Incremental Few-Shot Object Detection in Remote Sensing Images. [CoRR abs/2405.11293](https://dblp.org/db/journals/corr/corr2405.html#abs-2405-11293) (2024)
由于论文比较多,看得没那么仔细,在持续更新当中,如若有错误问题请指出!
一、背景:
二、定义:
设定:
01 模型可以在一个仅含有基类数据集上进行预训练,每个基类都有丰富的训练样本可用。(使用现有的数据集来引导模型是很有价值的)
02 iFSD模型一但训练之后,就具备部署到真实应用场景的能力,并且仅含有少量标注样 本的novel类可以随时进行注册(被检测)。(模型能保证在目前已见过的所有类上都具有良好的性能,没有遗忘)
03 从无限的例子流中学习新类在内存占用、存储和计算成本方面应该是可行的。理想情况下,该模型应支持在机器人和智能手机等资源有限的设备上进行部署。
增量小样本目标检测如何工作的:
数据集划分:
三、相关论文:
按照年份来进行统计
按照在Novel类上是否微调进行统计
01 Incremental Few-Shot Object Detection.**CVPR 2020: 13843-13852
论文地址 | 论文代码 | 是否微调 | 本地论文 |
---|---|---|---|
论文地址 | 论文代码 | 👍 | [本地论文](论文\增量小样本目标检测\2020\Incremental Few-Shot Object Detection (CVPR 2020).pdf) |
首次提出增量小样本目标检测这一方向,并且规定了其设定,如定义所示。 新类无需微调
02 Towards Generalized and Incremental Few-Shot Object Detection.** CoRR abs/2109.11336 (2021)
论文地址 | 论文代码 | 是否微调 | 本地论文 |
---|---|---|---|
论文地址 | 未公开 | 👎 | [本地论文](论文\增量小样本目标检测\2021\Towards Generalized and Incremental Few-Shot Object Detection(CVPR 2021).pdf) |
采用Base和Novel双分支结构,遇到Novel类时,仅仅微调Dynamic Learning Branch
03 Few-Shot Batch Incremental Road Object Detection via Detector Fusion.** ICCVW 2021: 3063-3070
Anuj Tambwekar, Kshitij Agrawal, Anay Majee, Anbumani Subramanian
论文地址 | 论文代码 | 是否微调 | 本地论文 |
---|---|---|---|
论文地址 | 未公开 | 👎 | [本地论文](论文\增量小样本目标检测\2021\Few-Shot Batch Incremental Road Object Detection via Detector Fusion (ICCV Workshop 2021).pdf) |
也是采用双分支的思想,基于Few X这个方法,Novel类上存在微调
04 Class-Incremental Few-Shot Object Detection.** CoRR abs/2105.07637 (2021)
Pengyang Li, Yanan Li, Donghui Wang
论文地址 | 论文代码 | 是否微调 | 本地论文 |
---|---|---|---|
论文地址 | 未公开 | 👎 | [本地论文](论文\增量小样本目标检测\2021\Class-Incremental_Few-Shot_Object_Detection(CoRR abs2105.07637 2021).pdf) |
increment阶段时候使用的是base+novel的平衡集,访问了base类数据
05 Sylph: A Hypernetwork Framework for Incremental Few-shot Object Detection.**CVPR 2022: 9025-9035
Li Yin, Juan M. Perez-Rua, Kevin J. Liang:
论文地址 | 论文代码 | 是否微调 | 本地论文 |
---|---|---|---|
论文地址 | 论文代码 | 👍 | [本地论文](论文\增量小样本目标检测\2022\Sylph A Hypernetwork Framework for Incremental Few-shot Object Detection (CVPR 2022).pdf) |
meta-learning + fcos ,不需要对Novel类进行微调
06 Incremental Few-Shot Object Detection for Robotics.** ICRA 2022: 8447-8453
Yiting Li, Haiyue Zhu, Sichao Tian, Fan Feng, Jun Ma
, Chek Sing Teo, Cheng Xiang, Prahlad Vadakkepat, Tong Heng Lee:
论文地址 | 论文代码 | 是否微调 | 本地论文 |
---|---|---|---|
论文地址 | 未公开 | 👎 | [本地论文](论文\增量小样本目标检测\2022\Incremental_Few-Shot_Object_Detection_for_Robotics(ICRA 2022).pdf) |
有微调的
07 Meta-Learning-Based Incremental Few-Shot Object Detection.** IEEE Trans. Circuits Syst. Video Technol. 32(4): 2158-2169 (2022)
Meng Cheng, Hanli Wang
, Yu Long:
论文地址 | 论文代码 | 是否微调 | 本地论文 |
---|---|---|---|
论文地址 | 论文代码 | 👎 | [本地论文](论文\增量小样本目标检测\2022\Meta-Learning Based Incremental Few-Shot Object Detection (TCSVT 2022).pdf) |
这篇论文和ONCE那篇是一样的,就是修改了一个Meta-Learner,Novel类不需要微调
08 Incremental few-shot object detection via knowledge transfer.** Pattern Recognit. Lett. 156: 67-73 (2022)
Hangtao Feng, Lu Zhang, Xu Yang, Zhiyong Liu:
论文地址 | 论文代码 | 是否微调 | 本地论文 |
---|---|---|---|
论文地址 | 未公开 | 👎 | [本地论文](论文\增量小样本目标检测\2022\Incremental Few-Shot Object Detection via Knowledge Transfer (Pattern Recognition Letters 2022).pdf) |
Novel类上有微调的
09 Incremental Few-Shot Object Detection via Simple Fine-Tuning Approach.** ICRA 2023: 9289-9295
论文地址 | 论文代码 | 是否微调 | 本地论文 |
---|---|---|---|
论文地址 | 未公开,有地址 | 👎 | [本地论文](论文\增量小样本目标检测\2023\Incremental Few-Shot Object Detection via Simple Fine-Tuning Approach (ICRA 2023).pdf) |
有微调,只是仅仅在Novel类数据集上微调,没有访问Base类数据而已
10 Novel-Registrable Weights and Region-Level Contrastive Learning for Incremental Few-shot Object Detection.** ICONIP (11) 2023: 400-412
Shiyuan Tang, Hefei Mei, Heqian Qiu, Xinpeng Hao, Taijin Zhao, Benliu Qiu, Haoyang Cheng, Jian Jiao, Chuanyang Gong, Hongliang Li:
论文地址 | 论文代码 | 是否微调 | 本地论文 |
---|---|---|---|
论文地址 | 未公开 | 👎 | [本地论文](论文\增量小样本目标检测\2023\Novel-Registrable Weights and Region-Level Contrastive Learning for Incremental Few-shot Object Detection(CCIS 2023).pdf) (P420) |
Novel类上也有微调
11 Incremental-DETR: Incremental Few-Shot Object Detection via Self-Supervised Learning.** AAAI 2023: 543-551
Na Dong, Yongqiang Zhang, Mingli Ding, Gim Hee Lee:
论文地址 | 论文代码 | 是否微调 | 本地论文 |
---|---|---|---|
论文地址 | 未公开,有地址 | 👎 | [本地论文](论文\增量小样本目标检测\2023\Incremental-DETR Incremental Few-Shot Object Detection via Self-Supervised Learning (AAAI 2023).pdf) |
Novel类上有微调
12 Class Incremental Robotic Pick-and-Place via Incremental Few-Shot Object Detection.** IEEE Robotics Autom. Lett. 8(9): 5974-5981 (2023)
Jieren Deng, Haojian Zhang
, Jianhua Hu
, Xingxuan Zhang
, Yunkuan Wang
:
论文地址 | 论文代码 | 是否微调 | 本地论文 |
---|---|---|---|
论文地址 | 未公开 | 👎 | [本地论文](论文\增量小样本目标检测\2023\Class_Incremental_Robotic_Pick-and-Place_via_Incremental_Few-Shot_Object_Detection(IEEE Robotics Autom 2023).pdf) |
Transformer方法,采用Adapter-tuning,Novel类上有微调
13 . Incremental Few-Shot Object Detection with scale- and centerness-aware weight generation. Comput. Vis. Image Underst. 235: 103774 (2023)
Lu Zhang, Xu Yang, Lu Qi, Shaofeng Zeng
, Zhiyong Liu:
论文地址 | 论文代码 | 是否微调 | 本地论文 |
---|---|---|---|
论文地址 | 未公开 | 👎 | [本地论文](论文\增量小样本目标检测\2023\Incremental Few-Shot Object Detection with scale- and centerness-aware weight generation(CVIU 2023).pdf) |
Novel类上有微调
14 Few-Shot Incremental Object Detection in Aerial Imagery via Dual-Frequency Prompt.** IEEE Trans. Geosci. Remote. Sens. 62: 1-17 (2024)
Xiaonan Lu, Wenhui Diao
, Junxi Li
, Yidan Zhang
, Peijin Wang
, Xian Sun
, Kun Fu:
论文地址 | 论文代码 | 是否微调 | 本地论文 |
---|---|---|---|
论文地址 | 未公开 | 👎 | [本地论文](论文\增量小样本目标检测\2024\Few-Shot_Incremental_Object_Detection_in_Aerial_Imagery_via_Dual-Frequency_Prompt(TGRS 2024).pdf) |
采用多分枝来处理Base类和Novel类,Novel类上还是有微调的
15 An adaptive lightweight small object detection method for incremental few-shot scenarios of unmanned surface vehicles.** Eng. Appl. Artif. Intell. 133: 107989 (2024)
Bo Wang, Peng Jiang, Zhuoyan Liu, Yueming Li, Jian Cao, Ye Li:
论文地址 | 论文代码 | 是否微调 | 本地论文 |
---|---|---|---|
论文地址 | 未公开 | 👎 | [本地论文](论文\增量小样本目标检测\2024\An adaptive lightweight small object detection method for incremental few-shot scenarios of unmanned surface vehicles(EAoAF2024).pdf) |
太复杂了,没咋看明白
16 InfRS: Incremental Few-Shot Object Detection in Remote Sensing Images. CoRR abs/2405.11293 (2024)
Wuzhou Li, Jiawei Zhou, Xiang Li, Yi Cao, Guang Jin, Xuemin Zhang:
论文地址 | 论文代码 | 是否微调 | 本地论文 |
---|---|---|---|
论文地址 | 未公开 | 👎 | [本地论文](论文\增量小样本目标检测\2024\InfRS Incremental Few-shot Object Detection in Remote Sensing Images.pdf) |
---------------------------------------------------------: | :-----------------------------------------: | :------: | :----------------------------------------------------------: |
| 论文地址 | 未公开 | 👎 | [本地论文](论文\增量小样本目标检测\2024\InfRS Incremental Few-shot Object Detection in Remote Sensing Images.pdf) |
哈哈哈,最后一篇了,没看