Papers › AquaVision: Automating the detection of waste in water bodies using deep transfer learning

AquaVision: Automating the detection of waste in water bodies using deep transfer learning

18 Jul 2020Case Studies in Chemical and Environmental Engineering 2020 7archive 2025-07-28

Harsh Panwar, P.K.Gupta, Mohammad Khubeb Siddiquib, Ruben Morales-Menendez, Prakhar Bhardwaj, Sudhansh Sharma, Iqbal H.Sarker

Water pollution is one of the serious threats in the society. More than 8 million tons of plastic are dumped in the oceans each year. In addition to that beaches are littered by tourists and residents all around the world. It is no secret that the aquatic life ecosystem is at a risk and soon the ratio of plastic/waste to the marine life particulary fish will be 1:1. Hence, in this paper, we have proposed a dataset known as AquaTrash which is based on TACO data set. Further, we have applied proposed state-of-the-art deep learning-based object detection model known as AquaVision over AquaTrash dataset. Proposed model detects and classifies the different pollutants and harmful waste items floating in the oceans and on the seashores with mean Average Precision (mAP) of 0.8148. The propose method localizes the waste object that help in cleaning the water bodies and contributes to environment by maintaining the aquatic ecosystem.

PaperPDFCode

Code

Repository list and official/mentioned flags are the archive's, frozen 2025-07-28. Reachability, where shown, is from one Syntology probe window (2026-09-16 to 2026-09-18); repositories not probed show nothing. GitHub stars are not tracked.

Code Syntology ran Syntology

Not run by Syntology. Nothing on this page verifies that the listed code works.

Tasks

Object DetectionTransfer Learningobject-detection

Results from the paper archive 2025-07-28

TaskDatasetModelMetricValueRank at snapshotLeaderboardReport
Object Detection AquaTrash Aquavision mean average precision 0.8148 #1 of 1 Archive leaderboard report

Ranks are positions in the archive's leaderboards as they stood at the 2025-07-28 snapshot. Results published since then are not among these rows, so a rank here is not a current standing.

Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections