/WASSAN-Tree-Plantation

Primary LanguageJupyter Notebook

Tree Plantation Detection

Tree Detection using YOLO V8 - Old Dataset

Trained the YOLOv8n model on the above mentioned dataset, running training for 50 epochs

Metrics on training data:

  • Box loss: 1.3701
  • Class predictions loss: 0.6704
  • Detection & Localization loss: 1.0026
  • Precision: 0.90212
  • Recall: 0.93482
  • Mean Average Precision ( threshold 0.5): 0.96124
  • Mean Average Precision ( threshold 0.5 to 0.95): 0.54982

Metrics on validation data:

  • Box loss: 1.4979
  • Class predictions loss: 0.66644
  • Detection & Localization loss: 1.0533

Metrics on validation data:

Tree Detection using YOLO V8 - New Dataset

Trained the YOLOv8n model on the above mentioned dataset, running training for 50 epochs

Metrics on training data:

  • Box loss: 2.1599
  • Class predictions loss: 1.297
  • Detection & Localization loss: 1.049
  • Precision: 0.82403
  • Recall: 0.78679
  • Mean Average Precision ( threshold 0.5): 0.80289
  • Mean Average Precision ( threshold 0.5 to 0.95): 0.35833

Metrics on validation data:

  • Box loss: 2.0365
  • Class predictions loss: 1.955
  • Detection & Localization loss: 1.0089

Extended Dataset 1

Trained the YOLOv8n model on the above mentioned dataset, running training for 50 epochs

Metrics on training data:

  • Box loss: 1.4272
  • Class predictions loss: 0.68797
  • Detection & Localization loss: 1.0129
  • Precision: 0.87044
  • Recall: 0.9089
  • Mean Average Precision ( threshold 0.5): 0.93178
  • Mean Average Precision ( threshold 0.5 to 0.95): 0.51629

Metrics on validation data:

  • Box loss: 1.5471
  • Class predictions loss: 0.72492
  • Detection & Localization loss: 1.0358

Extended Dataset 2

Trained the YOLOv8n model on the above mentioned dataset, running training for 50 epochs

Metrics on training data:

  • Box loss: 1.441
  • Class predictions loss: 0.72355
  • Detection & Localization loss: 1.0187
  • Precision: 0.89621
  • Recall: 0.90416
  • Mean Average Precision ( threshold 0.5): 0.93917
  • Mean Average Precision ( threshold 0.5 to 0.95): 0.50592

Metrics on validation data:

  • Box loss: 1.5887
  • Class predictions loss: 0.71573
  • Detection & Localization loss: 1.0401

Extended Dataset 3

Trained the YOLOv8n model on the above mentioned dataset, running training for 50 epochs

Metrics on training data:

  • Box loss: 1.4048
  • Class predictions loss: 0.66397
  • Detection & Localization loss: 1.0151
  • Precision: 0.92109
  • Recall: 0.93271
  • Mean Average Precision ( threshold 0.5): 0.96445
  • Mean Average Precision ( threshold 0.5 to 0.95): 0.57651

Metrics on validation data:

  • Box loss: 1.3906
  • Class predictions loss: 0.62483
  • Detection & Localization loss: 1.0077