Research map / Detection and segmentation / CNN features
FloW: A Dataset and Benchmark for Floating Waste Detection in Inland Waters
Yuwei Cheng, Jiannan Zhu, Mengxin Jiang, Jie Fu, Changsong Pang, Peidong Wang, Kris Sankaran, Olawale Onabola, Yimin Liu, Dianbo Liu, Yoshua Bengio
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| use Cascade R-CNN instead of Faster R-CNN | detector · FloW-Img test set | mmAP: 0.184 -> 0.434 | Improved |
| use Cascade R-CNN instead of Faster R-CNN | detector · FloW-Img test set | FPS: 9.3 -> 3.9 | Hurt |
| use Cascade R-CNN instead of FPN | detector · FloW-Img test set | mmAP: 0.334 -> 0.434 | Improved |
| use YOLO-v3 instead of DSSD (one-stage) | detector · FloW-Img test set | mmAP: 0.275 -> 0.128 | Hurt |
| evaluate on FloW-Img small objects instead of UAV-BD small objects (Cascade R-CNN) | Cascade R-CNN · small objects | mmAP: 0.491 -> 0.293 | Hurt |
| evaluate on FloW-Img small objects instead of UAV-BD small objects (FPN) | FPN · small objects | mmAP: 0.481 -> 0.213 | Hurt |
| evaluate on FloW-Img small objects instead of UAV-BD small objects (DSSD) | DSSD · small objects | mmAP: 0.527 -> 0.121 | Hurt |
| train on UAV-BD, TACO or MJU-Waste instead of FloW-Img training data | Cascade R-CNN ResNet-101 · FloW-Img test set | mmAP: 0.058, 0.069, 0.032 | Hurt |
| use radar RDM images instead of RGB images with Cascade R-CNN | Cascade R-CNN · FloW-RI test set | AP50: 0.640 -> 0.898 | Improved |
| add radar point cloud to image (CRF-Net fusion) versus image only | Cascade R-CNN image baseline · FloW-RI test set | AP50: 0.640 -> 0.754 | Improved |
| add radar point cloud to image (Li et al. fusion) versus image only | Cascade R-CNN image baseline · FloW-RI test set | AP50: 0.640 -> 0.677 | Improved |
| use fusion CRF-Net instead of radar RDM image alone | Cascade R-CNN radar RDM · FloW-RI test set | AP50: 0.898 -> 0.754 | Hurt |
Closest papers by meaning
- GLH-Water: A Large-Scale Dataset for Global Surface Water Detection in Large-Size Very-High-Resolution Satellite Imagery (AAAI 2024 · similarity 0.80)
- LAqua: Laplacian Pyramids for Aquatic Segmentation (Student Abstract) (AAAI 2025 · similarity 0.80)
- RipVIS: Rip Currents Video Instance Segmentation Benchmark for Beach Monitoring and Safety (CVPR 2025 · similarity 0.80)
- UrbanWaste: In-the-Bin Dataset for Waste Disposal Inspection with Multi-Granularity Hierarchical Labels (AAAI 2025 · similarity 0.79)
- WaterDiffusion: Learning a Prior-involved Unrolling Diffusion for Joint Underwater Saliency Detection and Visual Restoration (AAAI 2025 · similarity 0.79)
- WebUOT-1M: Advancing Deep Underwater Object Tracking with A Million-Scale Benchmark (NeurIPS 2024 · similarity 0.79)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
