-
I am using yolov4-tiny.weights with yolov4-tiny-relu-new_coords.cfg and using yolov4-tiny-relu-new_coords-tpu.cfg convert to tflite with compile.
But the perform on the edgetpu is 60sec per image. …
-
Hi, thanks for your wonderful work. However, I have got very different results from cpu and edgetpu. In the following image, left one is the result using cpu, and the right one is the result using edg…
-
Hi, I encountered one bug while using the detector over video. I detects 'ghosts'. I noticed that it draw bounding boxes where people are not present. maybe it is something with the conversion to boun…
-
# YOLOv4 on Edge TPU | loliot
YOLOv4 on Edge TPU
[https://wiki.loliot.net/docs/lang/python/libraries/yolov4/python-yolov4-edge-tpu](https://wiki.loliot.net/docs/lang/python/libraries/yolov4/python-y…
-
Any benchmarks, please
-
Hi,
I followed https://wiki.loliot.net/docs/lang/python/libraries/yolov4/python-yolov4-edge-tpu/ to train a model with only one class.
If I use the original yolov4-tiny.cfg, the training works no…
-
First of all, thanks a lot for the great work/package/documentation!
I trained a YoloV4 Tiny model using this script:
```python
from tensorflow.keras import callbacks, optimizers
from yolov4.t…
-
i trained yolov3 and yolov4 on my custom data using ultralic version. however, the mAP of yolov4(mish)is lower than yolov3 and yolov4(relu). But the paper mentioned yolov4 with mish activation functio…
-
[English Version]
Hi, I found the problem of quantize and compile in Vitis AI 1.4.1 and Vitis AI 2.x respectively
First mention the part of Vitis AI 1.4.1 or 1.4
The environment and training mode…
-
# TensorFlow 2 YOLOv4 | loliot
YOLOv4 Implemented in Tensorflow 2.
[https://wiki.loliot.net/docs/lang/python/libraries/yolov4/python-yolov4-about](https://wiki.loliot.net/docs/lang/python/libraries/…