train YOLOX on COCO

README.md

@@ -105,7 +105,7 @@ python train.py --cuda -d coco --root path/to/COCO -v yolov1 -bs 16 --max_epoch
 | YOLOv3 | DarkNet-53    |  640  |  √   |  250  |        42.9            |       63.5       | [ckpt](https://github.com/yjh0410/PyTorch_YOLO_Tutorial/releases/download/yolo_tutorial_ckpt/yolov3_coco.pth) |
 | YOLOv4 | CSPDarkNet-53 |  640  |  √   |  250  |                        |                  |  |
 | YOLOv5 | CSPDarkNet-L  |  640  |  √   |  250  |                        |                  |  |
-| YOLOX  | CSPDarkNet-L  |  640  |  √   |  250  |                        |                  |  |
+| YOLOX  | CSPDarkNet-L  |  640  |  √   |  300  |                        |                  |  |
 | YOLOv7 | ELANNet-Large |  640  |  √   |  300  |                        |                  |  |
 
 *All models are trained with ImageNet pretrained weight (IP). All FLOPs are measured with a 640x640 image size on COCO val2017. The FPS is measured with batch size 1 on 3090 GPU from the model inference to the NMS operation.*

train.sh

@@ -1,13 +1,13 @@
 # Train YOLO on VOC
 python train.py \
         --cuda \
-        -d voc \
+        -d coco \
         --root /mnt/share/ssd2/dataset/ \
         -m yolox \
         -bs 16 \
         -size 640 \
         --wp_epoch 1 \
-        --max_epoch 150 \
+        --max_epoch 300 \
         --eval_epoch 10 \
         --ema \
         --fp16 \