Blog
Blog
How to Train a YOLOv4 Tiny model and Use TensorFlow Lite
Using the state of the art YOLOv4 tiny Darknet model converted to TensorFlow Lite for on-device inference.
No items found.
YOLOv4 was a real-time object detection model published in April 2020 that achieved state-of-the-art performance on the COCO dataset. It works by breaking the object detection task into two pieces, regression to identify object positioning via bounding boxes and classification to determine the object's class. This implementation of YoloV4 uses the Darknet framework.
By using YOLOv4, you are implementing many of the past research contributions in the YOLO family along with a series of new contributions unique to YOLOv4 including new features: WRC, CSP, CmBN, SAT, Mish activation, Mosaic data augmentation, CmBN, DropBlock regularization, and CIoU loss. In short, with YOLOv4, you're using a better object detection network architecture and new data augmentation techniques.
As can be seen in the results below, YOLOv4 has an incredibly high performance for a very high FPS; this was a major improvement from previous object detection models which only had either high performance or high inference speeds.
Breaking Down YOLOv4: https://blog.roboflow.com/a-thorough-breakdown-of-yolov4/
Training YOLOv4 on a Custom Dataset: https://blog.roboflow.com/training-yolov4-on-a-custom-dataset/
YOLOv4 Darknet
is licensed under a
YOLO
license.
You can use Roboflow Inference to deploy a
YOLOv4 Darknet
API on your hardware. You can deploy the model on CPU (i.e. Raspberry Pi, AI PCs) and GPU devices (i.e. NVIDIA Jetson, NVIDIA T4).
Below are instructions on how to deploy your own model API.
