From 67e8683b22ca88fb2d9c865a4ac9ccf597b77578 Mon Sep 17 00:00:00 2001
From: yjh0410 <1394571815@qq.com>
Date: Tue, 8 Feb 2022 09:16:38 +0800
Subject: [PATCH] upload YOLOX backbone

---
 README.md | 4 ++++
 1 file changed, 4 insertions(+)

diff --git a/README.md b/README.md
index 3c1010c..3d1e015 100644
--- a/README.md
+++ b/README.md
@@ -10,6 +10,10 @@ In this project, you can enjoy:
 - CSPDarkNet-Tiny
 - CSPDarkNet-Nano
 
+# Requirements
+- Torch >= 1.9.1
+
+# Motivation
 As far as I know, it is difficult for many researchers to train YOLOv5 or YOLOX from scratch because they, including me, do not have sufficient computing resources. Therefore, a pre-trained backbone is still important. 
 
 In addition, since YOLOv5 and YOLOX are excellent object detectors, their backbones theoretically perform better than imagenet pretrained weights.