Conflict-Alleviated Gradient Descent for Adaptive Object Detection

Authors: Wenxu Shi, Bochuan Zheng

IJCAI 2024 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We further validate our theoretical analysis and methods on several DAOD tasks, including cross-camera, weather, scene, and synthetic-to-real-world adaptation. Extensive experiments on multiple DAOD benchmarks demonstrate the effectiveness and superiority of our CAGrad approach.
Researcher Affiliation Academia 1School of Microelectronics and Communication Engineering, Chongqing University, China 2School of Computer Science, China West Normal University, China wxshi@cqu.edu.cn, zhengbc@vip.163.com
Pseudocode Yes Algorithm 1 CAGrad+DAOD Optimization Algorithm
Open Source Code No The paper does not provide any explicit statement about releasing source code or a link to a code repository for the methodology described.
Open Datasets Yes Our methodology is rigorously evaluated on a variety of datasets, including KITTI [Geiger et al., 2012], Cityscapes [Cordts et al., 2016b], Foggy Cityscapes [Sakaridis et al., 2018b], Sim10k [Johnson-Roberson et al., 2016], and BDD100k [Yu et al., 2020]
Dataset Splits Yes Cityscapes as the source dataset, consisting of 2,975 training images and 500 evaluation images. [...] The BDD100k subset, encompassing 36,728 training and 5,258 validation images, is rich in diverse daylight scenes, each annotated with bounding boxes. [...] Sim10k [...] containing 10,000 training images with 58,701 bounding box annotations. Cityscapes is employed as the target domain, specifically focusing on car instances for both training and evaluation.
Hardware Specification Yes All these experiments are conducted using RTX3090 NVIDIA GPUs.
Software Dependencies No The paper mentions optimizers (SGD, Adam) and pre-trained models (ResNet50, VGG-16) but does not provide specific version numbers for any software dependencies like PyTorch, TensorFlow, or specific library versions.
Experiment Setup Yes Aligning with the Faster RCNN series, we train the network using the SGD optimizer with a momentum of 0.9 and a weight decay of 5 10 4. The initial learning rate is set to 1 10 3 and is reduced to 1 10 4 after 5 epochs. A total of 15 epochs are conducted, with a batch size of 2 maintained throughout. In line with the Deformable DETR series, we utilize the Adam optimizer [Kingma and Ba, 2015] for training over 50 epochs. The learning rate is initialized at 2 10 4 and reduced by a factor of 0.1 after 40 epochs. A batch size of 4 is employed consistently in all experiments.