Academic Journal
Reannotation-Free Automatic Recognition Methods for Empty Camera Trap Images Based on MegaDetector Optimization.
| Τίτλος: | Reannotation-Free Automatic Recognition Methods for Empty Camera Trap Images Based on MegaDetector Optimization. |
|---|---|
| Συγγραφείς: | Zhang, Mei1 (AUTHOR), Yao, Jing-Xu1,2 (AUTHOR), Wu, Rong-Hai1,3 (AUTHOR), Li, Xiao-Wei1,4 (AUTHOR), Ren, Guo-Peng2,5 (AUTHOR), Xiao, Wen1,3,4,5 (AUTHOR), Yang, Deng-Qi1,2,3,4 (AUTHOR) dengqiyang@163.com |
| Πηγή: | Animals (2076-2615). Aug2026, Vol. 16 Issue 16, p2609. 16p. |
| Θεματικοί όροι: | *Error rates, *False positive error, *Scouting cameras, *Mathematical optimization, *Image recognition (Computer vision), *Wildlife monitoring |
| Περίληψη: | Simple Summary: Camera trapping produces massive amounts of empty images caused by misfires, which takes researchers plenty of time to sort out manually. MegaDetector is the mainstream tool to screen empty images, yet users struggle to pick an appropriate confidence threshold: a higher threshold misses more animal images, while a lower one retains too many useless empty shots, and it is hard to balance these two errors. To solve this practical problem in wildlife image processing, we designed three improved schemes built upon MegaDetector for different research demands. MD_CS targets researchers who want to avoid missing wildlife records, cutting missed detections via context similarity comparison. MD_CL fits users prioritizing fewer empty images, utilizing machine learning to lower false triggers. The combined MD_CS_CL method delivers the best overall performance for those pursuing low error rates on both sides. Tests on multiple wildlife datasets prove all three approaches outperform simple threshold tuning. They reach ideal trade-offs between missed animal images and false empty frames, run efficiently on regular GPUs, and eliminate manual labeling work. This toolkit effectively cuts screening workload and offers a practical technical solution for large-scale wildlife camera trapping surveys. Camera traps are widely used in wildlife surveys but often generate numerous empty images due to false triggers, which are time-consuming to filter manually. Microsoft's MegaDetector (MD) is commonly applied for empty image filtering; however, setting its confidence threshold is challenging because a high threshold increases the false negative rate (FNR), whereas a low threshold raises the false positive rate (FPR), making it difficult to achieve a satisfactory trade-off between the two. To address this dilemma, we proposed three MD-based methods with different performance and computational requirements. For FNR-sensitive users, we proposed MD_CS, which integrated context image similarity to achieve a very low FNR while controlling the FPR. For FPR-sensitive users, we proposed MD_CL, which used MD to generate supervisory signals for training a classification model, thereby reducing the FPR while controlling the FNR. For users requiring both low FNR and low FPR, we proposed MD_CS_CL, an ensemble method that combined the strengths of MD_CS and MD_CL. Experimental results on multiple datasets showed that these three methods achieved a significantly better balance between FNR and FPR compared to the threshold adjustment method of the MD method. Specifically, MD_CS maintained an FNR of 1.8–5.9% while controlling the FPR at 4.9–8.5%; MD_CL kept the FPR at 0.4–2.6% while controlling the FNR at 3.8–8.4%; MD_CS_CL achieved an optimal balance with an FNR of 1.4–4.0% and an FPR of 0.3–0.9%. On a server equipped with an RTX 4090 GPU, the three methods processed 6.7, 5.4, and 4.0 images per second, respectively, and none required manual annotation. Consequently, the proposed methods reduce labor costs, improve work efficiency, and provide reliable technical support for large-scale wildlife monitoring by achieving a superior FNR–FPR trade-off compared to MD's traditional threshold regulation. [ABSTRACT FROM AUTHOR] |
| Βάση Δεδομένων: | Academic Search Index |
καταχωρήστε σχόλιο πρώτοι!