目录
- 一，准备
-
- 1.1代码
- 1.2 环境准备
- 1.3 搭建镜像
- 1.4 配置slowfast环境
- 1.5 ava.json
- 1.6 SLOWFAST_32x2_R101_50_50.yaml
- 1.7 SLOWFAST_32x2_R101_50_50 .pkl
- 二，代码运行
- 三 错误解决
复现过程视频：
B站复现视频
复现结果
一，准备
1.1代码
SlowFast官网地址
代码下载：
git clone https://github.com/facebookresearch/slowfast
这里建议使用码云来下载，使用上面的那个命令实在太慢，码云下载方法：
git clone 显著提速，解决Github代码拉取速度缓慢问题
下载链接
1.2 环境准备
这里推荐几个网站，租用这些网站的GPU然后跑代码，因为普通的笔记本搭建图像处理的环境会遇到各种各样的问题。在网站上租用GPU可以使用官方搭建好的镜像，如果跑代码或者自己把环境弄乱了，就可以释放实例，重新再来（约等于自己电脑的重装系统，但是这个租用的释放和搭建新镜像速度非常快）
以下是我推荐的几个网站：
1，极链AI云平台（我用的就是这个，有50元的白嫖额度，我就是靠这50元的额度跑出来slowfast）
2，MistGPU
1.3 搭建镜像
第一步，在极链AI云平台选择一个机子，我选的一般都是4元/小时的。
第二步，选择GPU数量，我们跑自己的视频，一般1个GPU就够了
第三步，实例镜像，官方给的框架是PyTorch，这里也就PyTorch，PyTorch的版本官方给的1.3（当时官方发布的时候，PyTorch最高版本就1.3，但是这里的镜像没有1.3，所以选择一个比它高的1.4版本）。注意，python一定要选3.7版本，python一定要选3.7版本，python一定要选3.7版本，3.6版本我是跑出了问题。CUDA只有10.1可选（这也是官方镜像的好处，在自己电脑上安装CUDA很烦的）
备注：有读者尝试的时候，这里出现问题，这是我2021年2月份做出来的，读者是4月份尝试的，然后报错，这里读者采用了pytorch1.4的版本，cuda10.2的版本。所以如果版本问题的话，尝试把pytorch和cuda版本提高。
第四步，按照极链AI云平台的手册指引，在Xshell上打开实例。
1.4 配置slowfast环境
按照官网上的步骤来
Installation
注意，这里的opencv-python就不要执行，因为镜像里自带了opencv-python！
1.5 ava.json
{"bend/bow (at the waist)": 0, "crawl": 1, "crouch/kneel": 2, "dance": 3, "fall down": 4, "get up": 5, "jump/leap": 6, "lie/sleep": 7, "martial art": 8, "run/jog": 9, "sit": 10, "stand": 11, "swim": 12, "walk": 13, "answer phone": 14, "brush teeth": 15, "carry/hold (an object)": 16, "catch (an object)": 17, "chop": 18, "climb (e.g., a mountain)": 19, "clink glass": 20, "close (e.g., a door, a box)": 21, "cook": 22, "cut": 23, "dig": 24, "dress/put on clothing": 25, "drink": 26, "drive (e.g., a car, a truck)": 27, "eat": 28, "enter": 29, "exit": 30, "extract": 31, "fishing": 32, "hit (an object)": 33, "kick (an object)": 34, "lift/pick up": 35, "listen (e.g., to music)": 36, "open (e.g., a window, a car door)": 37, "paint": 38, "play board game": 39, "play musical instrument": 40, "play with pets": 41, "point to (an object)": 42, "press": 43, "pull (an object)": 44, "push (an object)": 45, "put down": 46, "read": 47, "ride (e.g., a bike, a car, a horse)": 48, "row boat": 49, "sail boat": 50, "shoot": 51, "shovel": 52, "smoke": 53, "stir": 54, "take a photo": 55, "text on/look at a cellphone": 56, "throw": 57, "touch (an object)": 58, "turn (e.g., a screwdriver)": 59, "watch (e.g., TV)": 60, "work on a computer": 61, "write": 62, "fight/hit (a person)": 63, "give/serve (an object) to (a person)": 64, "grab (a person)": 65, "hand clap": 66, "hand shake": 67, "hand wave": 68, "hug (a person)": 69, "kick (a person)": 70, "kiss (a person)": 71, "lift (a person)": 72, "listen to (a person)": 73, "play with kids": 74, "push (another person)": 75, "sing to (e.g., self, a person, a group)": 76, "take (an object) from (a person)": 77, "talk to (e.g., self, a person, a group)": 78, "watch (a person)": 79}
将上面的内容，保存在ava.json里面，然后保存在/home/slowfast/demo/AVA目录下。
1.6 SLOWFAST_32x2_R101_50_50.yaml
在目录：/home/slowfast/demo/AVA中打开SLOWFAST_32x2_R101_50_50.yaml
内容改为如下：
TRAIN:ENABLE: FalseDATASET: avaBATCH_SIZE: 16EVAL_PERIOD: 1CHECKPOINT_PERIOD: 1AUTO_RESUME: TrueCHECKPOINT_FILE_PATH: '/home/slowfast/configs/AVA/c2/SLOWFAST_32x2_R101_50_50.pkl' #path to pretrain modelCHECKPOINT_TYPE: pytorch
DATA:NUM_FRAMES: 32SAMPLING_RATE: 2TRAIN_JITTER_SCALES: [256, 320]TRAIN_CROP_SIZE: 224TEST_CROP_SIZE: 256INPUT_CHANNEL_NUM: [3, 3]
DETECTION:ENABLE: TrueALIGNED: False
AVA:BGR: FalseDETECTION_SCORE_THRESH: 0.8TEST_PREDICT_BOX_LISTS: ["person_box_67091280_iou90/ava_detection_val_boxes_and_labels.csv"]
SLOWFAST:ALPHA: 4BETA_INV: 8FUSION_CONV_CHANNEL_RATIO: 2FUSION_KERNEL_SZ: 5
RESNET:ZERO_INIT_FINAL_BN: TrueWIDTH_PER_GROUP: 64NUM_GROUPS: 1DEPTH: 101TRANS_FUNC: bottleneck_transformSTRIDE_1X1: FalseNUM_BLOCK_TEMP_KERNEL: [[3, 3], [4, 4], [6, 6], [3, 3]]SPATIAL_DILATIONS: [[1, 1], [1, 1], [1, 1], [2, 2]]SPATIAL_STRIDES: [[1, 1], [2, 2], [2, 2], [1, 1]]
NONLOCAL:LOCATION: [[[], []], [[], []], [[6, 13, 20], []], [[], []]]GROUP: [[1, 1], [1, 1], [1, 1], [1, 1]]INSTANTIATION: dot_productPOOL: [[[2, 2, 2], [2, 2, 2]], [[2, 2, 2], [2, 2, 2]], [[2, 2, 2], [2, 2, 2]], [[2, 2, 2], [2, 2, 2]]]
BN:USE_PRECISE_STATS: FalseNUM_BATCHES_PRECISE: 200
SOLVER:MOMENTUM: 0.9WEIGHT_DECAY: 1e-7OPTIMIZING_METHOD: sgd
MODEL:NUM_CLASSES: 80ARCH: slowfastMODEL_NAME: SlowFastLOSS_FUNC: bceDROPOUT_RATE: 0.5HEAD_ACT: sigmoid
TEST:ENABLE: FalseDATASET: avaBATCH_SIZE: 8
DATA_LOADER:NUM_WORKERS: 2PIN_MEMORY: TrueNUM_GPUS: 1
NUM_SHARDS: 1
RNG_SEED: 0
OUTPUT_DIR: .
#TENSORBOARD:
# MODEL_VIS:
# TOPK: 2
DEMO:ENABLE: TrueLABEL_FILE_PATH: "/home/slowfast/demo/AVA/ava.json"INPUT_VIDEO: "/home/slowfast/Vinput/2.mp4"OUTPUT_FILE: "/home/slowfast/Voutput/1.mp4"DETECTRON2_CFG: "COCO-Detection/faster_rcnn_R_50_FPN_3x.yaml"DETECTRON2_WEIGHTS: detectron2://COCO-Detection/faster_rcnn_R_50_FPN_3x/137849458/model_final_280758.pkl
下面这一行代码存放的是输入视频的位置，当然Vinput这个文件夹是我自己建的
INPUT_VIDEO: "/home/slowfast/Vinput/2.mp4"
下面这一行代码存放的是检测后视频的位置，当然Voutput这个文件夹是我自己建的
OUTPUT_FILE: "/home/slowfast/Voutput/1.mp4"
1.7 SLOWFAST_32x2_R101_50_50 .pkl
下载页面
下载模型SLOWFAST_32x2_R101_50_50 .pkl 到/home/slowfast/configs/AVA/c2目录下
然后在目录：/home/slowfast/demo/AVA下文件SLOWFAST_32x2_R101_50_50.yaml里修改如下：
TRAIN:ENABLE: FalseDATASET: avaBATCH_SIZE: 16EVAL_PERIOD: 1CHECKPOINT_PERIOD: 1AUTO_RESUME: TrueCHECKPOINT_FILE_PATH: '/home/slowfast/configs/AVA/c2/SLOWFAST_32x2_R101_50_50.pkl' #path to pretrain modelCHECKPOINT_TYPE: pytorch
二，代码运行
在xShell终端中，跳转到以下目录
cd /home/slowfast/
输入运行命令
python tools/run_net.py --cfg demo/AVA/SLOWFAST_32x2_R101_50_50.yaml
最后结果：
最后结果上传到了B站上：视频链接
三 错误解决
from av._core import time_base, pyav_version as version, library_versions
ImportError: libopenh264.so.5: cannot open shared object file: No such file or directory
解决方案：
参考之前写的解决方案链接
直接再终端运行：
conda install x264=='1!152.20180717' ffmpeg=4.0.2 -c conda-forge
任何程序错误，以及技术疑问或需要解答的，请添加
