Statements (58)
| Predicate | Object |
|---|---|
| gptkbp:instance_of |
gptkb:cosmic_ray_detector
|
| gptkbp:application |
gptkb:robotics
Surveillance Autonomous driving Video analysis |
| gptkbp:architecture |
gptkb:neural_networks
|
| gptkbp:based_on |
gptkb:YOLOv4
|
| gptkbp:community_support |
User forums
Tutorials available Active Git Hub repository |
| gptkbp:developed_by |
gptkb:Joseph_Redmon
|
| gptkbp:evaluates |
Custom datasets
Benchmark datasets m AP (mean Average Precision) FPS (Frames Per Second) |
| gptkbp:features |
High speed
Good accuracy Lightweight model |
| https://www.w3.org/2000/01/rdf-schema#label |
YOLOv4-tiny
|
| gptkbp:input_output |
RGB images
416x416 pixels Bounding boxes and class probabilities Class labels Confidence scores JSON or CSV |
| gptkbp:is_a_framework_for |
gptkb:Deep_Learning
|
| gptkbp:is_compared_to |
gptkb:Faster_R-CNN
gptkb:YOLOv3 gptkb:SSD |
| gptkbp:is_optimized_for |
Real-time object detection
|
| gptkbp:license |
GPL-3.0
|
| gptkbp:model |
6.9 MB
|
| gptkbp:performance |
High precision
Real-time performance High recall Faster than YOLOv4 |
| gptkbp:provides_information_on |
gptkb:COCO_dataset
|
| gptkbp:release_date |
gptkb:2020
|
| gptkbp:requires |
NVIDIA GPU for optimal performance
|
| gptkbp:successor |
gptkb:YOLOv5
|
| gptkbp:supports |
ONNX format
Tensor RT optimization Multi-class detection Multi-object detection |
| gptkbp:training |
Fine-tuning on specific datasets
Shorter than YOLOv4 Transfer learning from YOLOv4 |
| gptkbp:uses |
Data augmentation
Transfer learning Batch normalization Skip connections Darknet framework Leaky Re LU activation Dropout layers Data preprocessing techniques |
| gptkbp:written_in |
C and CUDA
|
| gptkbp:bfsParent |
gptkb:YOLO_family
|
| gptkbp:bfsLayer |
7
|