Weakly-Supervised Camouflaged Object Detection with Scribble Annotations

Authors: Ruozhen He, Qihua Dong, Jiaying Lin, Rynson W.H. Lau

AAAI 2023 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Experimental results show that our model outperforms relevant SOTA methods on three COD benchmarks with an average improvement of 11.0% on MAE, 3.2% on S-measure, 2.5% on E-measure, and 4.4% on weighted F-measure.
Researcher Affiliation Academia Department of Computer Science, City University of Hong Kong {ruozhenhe2-c, qihuadong2-c, jiayinlin5-c}@my.cityu.edu.hk, Rynson.Lau@cityu.edu.hk
Pseudocode No The paper does not contain any structured pseudocode or algorithm blocks.
Open Source Code Yes 1The code and dataset are available at https://github.com/ dddraxxx/Weakly-Supervised-Camouflaged-Object-Detectionwith-Scribble-Annotations.
Open Datasets Yes Our experiments are conducted on three COD benchmarks, CAMO(Le et al. 2019), CHAMELEON(Skurowski et al. 2018), and COD10K(Fan et al. 2020a). Following previous studies, we relabel 4,040 images (3,040 from COD10K, 1,000 from CAMO) and propose the S-COD dataset for training. The remaining is for testing.
Dataset Splits No The paper mentions "training set" and "testing" but does not explicitly specify a validation set or its split details.
Hardware Specification Yes We implement our method with Py Torch and conduct experiments on a Ge Force RTX2080Ti GPU.
Software Dependencies No The paper mentions "Py Torch" but does not provide specific version numbers for software dependencies.
Experiment Setup Yes In the training phase, input images are resized to 320 320 with horizontal flips. We use the stochastic gradient descent (SGD) optimizer with a momentum of 0.9, a weight decay of 5e-4, and triangle learning rate schedule with maximum learning rate 1e-3. The batch size is 16, and the training epoch is 150.