adaptive fusion of faster r-cnn and shape from shading  (IEEE Access)

Journal: Sensors (Basel, Switzerland)

Article Title: Evaluation of Automated Object-Detection Algorithms for Koala Detection in Infrared Aerial Imagery

doi: 10.3390/s24217048

Figure Lengend Snippet: Parameter settings of different methods.

Article Snippet: This slow speed was primarily due to the small size of the koalas and hence the need to run multiple tiles through the detector for each image. shows the average processing times for MATLAB implementations of the AAGD, IAAGD, HB-MLCM, ILCM, MLCM, MPCM, TMBM, Faster R-CNN, YOLOv2, and Combined 2DCNN methods.

Techniques:

Journal: Sensors (Basel, Switzerland)

Article Title: Evaluation of Automated Object-Detection Algorithms for Koala Detection in Infrared Aerial Imagery

doi: 10.3390/s24217048

Figure Lengend Snippet: Evaluation curves for 11 comparative koala detection techniques (AAGD, IAAGD, HB-MLCM, ILCM, MLCM, MPCM, TMBM, Faster R-CNN, YOLOv2, Combined 2DCNN, and the MOBIVLS): ( a 1 – d 1 ) show the receiver operating characteristic ( R O C ) curves (TPR vs. FPR); ( a 2 – d 2 ) show the recall vs. (1-precision) curves; and ( a 3 – d 3 ) show the A U R O C and E E R percentages. The F P R range over which the A U R O C calculations were computed was (0– 10 − 4 ), while T P R range used was (0–1). The uppermost three rows of Figures show the results from datasets A–C, respectively, with the last row showing the overall (average) results. In all cases, the proposed MOBIVLS algorithm outperformed all of the other approaches tested.

Article Snippet: This slow speed was primarily due to the small size of the koalas and hence the need to run multiple tiles through the detector for each image. shows the average processing times for MATLAB implementations of the AAGD, IAAGD, HB-MLCM, ILCM, MLCM, MPCM, TMBM, Faster R-CNN, YOLOv2, and Combined 2DCNN methods.

Techniques:

Journal: Sensors (Basel, Switzerland)

Article Title: Evaluation of Automated Object-Detection Algorithms for Koala Detection in Infrared Aerial Imagery

doi: 10.3390/s24217048

Figure Lengend Snippet: The overall results of the three datasets, which were computed by treating the datasets as a single entity. Several performance metrics were computed for different object-detection techniques at (a) F P R of 10 − 6 and (b) 10 − 5 . The total number of unique koalas was 56. The best result of each metric is highlighted by an underline and bold style. The second-best result is indicated by bold style only. The proposed MOBIVLS algorithm performed better than all of the other techniques against all of the metrics used. The column ‘Time’ represents the processing time in seconds per frame.

Article Snippet: This slow speed was primarily due to the small size of the koalas and hence the need to run multiple tiles through the detector for each image. shows the average processing times for MATLAB implementations of the AAGD, IAAGD, HB-MLCM, ILCM, MLCM, MPCM, TMBM, Faster R-CNN, YOLOv2, and Combined 2DCNN methods.

Techniques:

Journal: Sensors (Basel, Switzerland)

Article Title: Evaluation of Automated Object-Detection Algorithms for Koala Detection in Infrared Aerial Imagery

doi: 10.3390/s24217048

Figure Lengend Snippet: Results from Dataset A of several performance metrics for different object-detection techniques at (a) F P R of 10 − 6 and (b) 10 − 5 . The total number of unique koalas was 25. The best result of each metric is highlighted by an underline and bold style. The second-best result is written in bold style only. The proposed MOBIVLS algorithm performed better than all of the other techniques against all of the metrics used.

Article Snippet: This slow speed was primarily due to the small size of the koalas and hence the need to run multiple tiles through the detector for each image. shows the average processing times for MATLAB implementations of the AAGD, IAAGD, HB-MLCM, ILCM, MLCM, MPCM, TMBM, Faster R-CNN, YOLOv2, and Combined 2DCNN methods.

Techniques:

Journal: Sensors (Basel, Switzerland)

Article Title: Evaluation of Automated Object-Detection Algorithms for Koala Detection in Infrared Aerial Imagery

doi: 10.3390/s24217048

Figure Lengend Snippet: Results from Dataset B of several performance metrics for different object-detection techniques at (a) F P R of 10 − 6 and (b) 10 − 5 . The total number of unique koalas was 25. The best result of each metric is highlighted by an underline and bold style. The second-best result is written in bold style only. The proposed MOBIVLS algorithm performed better than all of the other techniques against all of the metrics used.

Article Snippet: This slow speed was primarily due to the small size of the koalas and hence the need to run multiple tiles through the detector for each image. shows the average processing times for MATLAB implementations of the AAGD, IAAGD, HB-MLCM, ILCM, MLCM, MPCM, TMBM, Faster R-CNN, YOLOv2, and Combined 2DCNN methods.

Techniques:

Journal: Sensors (Basel, Switzerland)

Article Title: Evaluation of Automated Object-Detection Algorithms for Koala Detection in Infrared Aerial Imagery

doi: 10.3390/s24217048

Figure Lengend Snippet: Results from Dataset C of several performance metrics for different object-detection techniques at (a) F P R of 10 − 6 and (b) 10 − 5 . The total number of unique koalas was 6. The best result of each metric is highlighted by an underline and bold style. The second-best result is written in bold style only. The proposed MOBIVLS algorithm performed better than all of the other techniques against all of the metrics used.

Article Snippet: This slow speed was primarily due to the small size of the koalas and hence the need to run multiple tiles through the detector for each image. shows the average processing times for MATLAB implementations of the AAGD, IAAGD, HB-MLCM, ILCM, MLCM, MPCM, TMBM, Faster R-CNN, YOLOv2, and Combined 2DCNN methods.

Techniques:

Journal: Frontiers in Artificial Intelligence

Article Title: A review on artificial intelligence for the diagnosis of fractures in facial trauma imaging

doi: 10.3389/frai.2023.1278529

Figure Lengend Snippet: Chronological order of applications and developments of AI for facial fracture detection reviewed in this study.

Article Snippet: In the context of adult head injuries, an approach for the detection of skull fractures was developed, building upon the Faster R-CNN framework (Kuang et al., ).

Techniques:

Journal: Mayo Clinic Proceedings: Digital Health

Article Title: A Review of Computer-Aided Diagnostic Algorithms for Cervical Neoplasia and an Assessment of Their Applicability to Female Genital Schistosomiasis

doi: 10.1016/j.mcpdig.2023.04.007

Figure Lengend Snippet: Summary of Eligible Cervical Cancer CAD Algorithms a , b , c , d

Article Snippet: Elakkiya et al, 2021 , 99.49 , 96.92 , 98.55 , Faster R-CNN + GAN , 1,112 positives; 1,993 negatives , 80:20 , Kaggle challenge, clinical study.

Techniques: