fix(cv demo): fix README bug

Change-Id: I6a9ec90e24a4b6d03bdd3d6ff374bce998d32c46
sophgo · Sep 12, 2024 · a769d83 · a769d83
1 parent 56bddfc
commit a769d83
Show file tree

Hide file tree

Showing 8 changed files with 22 additions and 19 deletions.
diff --git a/element/multimedia/decode/README.md b/element/multimedia/decode/README.md
@@ -5,7 +5,7 @@
 sophon-stream decode element是sophon-stream框架中的一个插件，用于图片、视频、RTSP/RTMP/GB28181视频流解码，以供后续的分析和处理使用。
 
 ## 1. 特性
-* 支持多种输入格式，如RTSP、RTMP、GB28181、本地视频、图片文件、BASE64等。
+* 支持多种输入格式，如RTSP、RTMP、GB28181、本地视频、图片文件、BASE64、CAMERA等。
 * 支持RTSP/RTMP/GB28181视频流断开重连。
 * 支持本地视频与图片文件配置循环。
 * 支持多路视频流高性能解码，支持硬件加速。
@@ -81,7 +81,7 @@ sophon-stream解码器插件具有一些可配置的参数，可以根据需求
 |:-------------:| :-------: | :------------------:| :------------------------:|
 | channel_id | 整数   | 无 | 输入数据通道编号 |
 |   url      | 字符串 | 无 | 输入数据路径，包括本地视频、图片、视频流和base64对应url后缀 |
-|source_type | 字符串  | 无  | 输入数据类型，"RTSP"代表RTSP视频流，“RTMP”代表RTMP视频流，“GB28181”代表GB28181视频流，“VIDEO”代表本地视频，“IMG_DIR”代表图片文件夹， “BASE64”代表base64数据 |
+|source_type | 字符串  | 无  | 输入数据类型，"RTSP"代表RTSP视频流，“RTMP”代表RTMP视频流，“GB28181”代表GB28181视频流，“VIDEO”代表本地视频，“IMG_DIR”代表图片文件夹， “BASE64”代表base64数据 , “CAMERA”摄像头或者其他类型的视频输入设备|
 |sample_interval | 整数  | 1  |抽帧数，如设置为5，表示每5帧有1帧会被后续处理，即为ObjectMata mFilter字段为false|
 |loop_num | 整数  | 1  | 循环次数，仅适用于source_type为"VIDEO"和“IMG_DIR”，值为0时无限循环|
 |fps | 浮点数  | 30 | 用于控制视频流的fps，fps=-1表示不控制fps；其它情况下，source_type为"IMG_DIR"或"BASE64"时由设置的值决定，其他source_type从视频流读取fps，设置的值不生效|
@@ -110,3 +110,4 @@ IMG_DIR/
 >2. 输入RTMP数据流的URL须以`rtmp://`开头
 >3. 假设输入BASE64的URL为`/base64`，则http请求的格式需为「POST」(http://{host_ip}:{base64_port}/base64)，request body的data字段存储base64数据，如{"data": "{base64 string，不含头部(data:image/xxx;base64,)}"}
 >4. 输入GB28181数据流的URL须以`gb28181://`开头
+>5. 输入CAMERA数据流的URL须以`/dev/video`开头
diff --git a/element/multimedia/decode/README_EN.md b/element/multimedia/decode/README_EN.md
@@ -5,7 +5,7 @@ English | [简体中文](README.md)
 The sophon-stream decode element is a plugin within the sophon-stream framework, used for decoding images, videos, RTSP/RTMP video streams for subsequent analysis and processing purposes.
 
 ## 1. feature
-* Supports various input formats, including RTSP, RTMP, local videos, image files, BASE64, etc.
+* Supports various input formats, including RTSP, RTMP, local videos, image files, BASE64, CAMERA ,etc.
 * Supports reconnection for interrupted RTSP/RTMP video streams.
 * Allows configuration for looping local videos and image files.
 * High-performance decoding for multiple video streams with hardware acceleration.
@@ -81,7 +81,7 @@ Additionally, attention should be paid to the setting of the input data channels
 |:-------------:| :-------: | :------------------:| :------------------------:|
 | channel_id | int   | \ | Input data channel number |
 |   url      | string | \ | Input data path, including local videos, images, video streams, and base64-encoded URLs. |
-|source_type | string  | \  | Input data types: "RTSP" represents an RTSP video stream, “RTMP” represents an RTMP video stream, “VIDEO” represents local videos, “IMG_DIR” represents image folders, and “BASE64” represents base64-encoded data. |
+|source_type | string  | \  | Input data types: "RTSP" represents an RTSP video stream, “RTMP” represents an RTMP video stream, “VIDEO” represents local videos, “IMG_DIR” represents image folders, “BASE64” represents base64-encoded data, and “CAMERA” camera or other type of video input device. |
 |sample_interval | int  | 1  |Frame extraction rate. Setting it to 5 implies that for every 5 frames, 1 frame will be processed subsequently, which means the ObjectMata mFilter field is set to false.|
 |loop_num | int  | 1  | Loop count. Only applicable when the source_type is set to "VIDEO" and "IMG_DIR". A value of 0 indicates an infinite loop.|
 |fps | float  | 30 | Used to control the frames per second (fps) of the video stream. Fps=-1 means no control over fps. In other cases, when source_type is set to "IMG_DIR" or "BASE64", it's determined by the set value. For other source_types, fps is read from the video stream, and the set value does not take effect.|
@@ -110,4 +110,4 @@ IMG_DIR/
 >2. The URL for inputting RTMP data stream must begin with `rtmp://`.
 >3. If the input BASE64 URL is `/base64`, the HTTP request format should be a POST request to "http://{host_ip}:{base64_port}/base64". The request body's data field stores the base64 data, such as {"data": "{base64 string, excluding the header (data:image/xxx;base64,)}"}.
 >4. The URL for inputting GB28181 data stream must start with `gb28181://`.
-
+>5. The URL for inputting CAMERA data stream must start with `/dev/video`.
diff --git a/element/tools/dwa/src/dwa.cc b/element/tools/dwa/src/dwa.cc
@@ -68,8 +68,7 @@ common::ErrorCode Dwa::initInternal(const std::string& json) {
 
   dst_h = configure.find(CONFIG_INTERNAL_DST_H_FILED)->get<int>();
   dst_w = configure.find(CONFIG_INTERNAL_DST_W_FILED)->get<int>();
-  resize_h = configure.find(CONFIG_INTERNAL_RESIZE_H_FILED)->get<int>();
-  resize_w = configure.find(CONFIG_INTERNAL_RESIZE_W_FILED)->get<int>();
+
   auto dwa_mode_str =
       configure.find(CONFIG_INTERNAL_DWA_MODE_FILED)->get<std::string>();
   dwa_mode = dwa_mode_map[dwa_mode_str];
@@ -111,6 +110,8 @@ common::ErrorCode Dwa::initInternal(const std::string& json) {
 
   } else if (dwa_mode ==
              DWA_FISHEYE_MODE) {  // 用于04e10 blend 2240x2240 不需要resize
+    resize_h = configure.find(CONFIG_INTERNAL_RESIZE_H_FILED)->get<int>();
+    resize_w = configure.find(CONFIG_INTERNAL_RESIZE_W_FILED)->get<int>();
     fisheye_attr = {0};
     // set_fish_default_param(&fisheye_attr);
     rot_mode = BMCV_ROTATION_180;

diff --git a/samples/bird_dwa_blend_encode/README.md b/samples/bird_dwa_blend_encode/README.md
@@ -43,7 +43,7 @@ chmod -R +x scripts/
 ```
 
 脚本执行完毕后，会在当前目录下生成`data`目录，其中包含`models`和`videos`两个子目录。
-
+```bash
 .
 ├── gridinfo
 │   ├── 0grid_info_bev_60_60_2920_60_60_dst_1920x1920_src_1920x1080.dat
@@ -105,7 +105,7 @@ chmod -R +x scripts/
     ├── c01_alpha_444p_m2__0_3072x32.bin
     ├── c01_beta_444p_m2__0_2240x32.bin
     └── c01_beta_444p_m2__0_3072x32.bin
-
+```
 
 ## 4. 环境准备
 

diff --git a/samples/dwa_blend_encode/README.md b/samples/dwa_blend_encode/README.md
@@ -23,7 +23,8 @@
 
 本例程用于说明如何使用sophon-stream快速构建深度估计应用。
 
-本例程中，鱼眼拼接算法的鱼眼展开、鱼眼拼接分别在两个element上进行运算，element内部可以开启多个线程，保证了一定的运行效率。
+本例程中，鱼眼拼接算法的鱼眼展开、鱼眼拼接分别在两个element上进行运算，element内部可以开启多个线程，保证了一定的运行效率。下图是广角拼接应用的流程图：
+![dwa_pipeline](pic/image.jpg)
 
 ## 2. 特性
 
@@ -43,13 +44,13 @@ chmod -R +x scripts/
 ```
 
 脚本执行完毕后，会在当前目录下生成`data`目录，其中包含`models`和`videos`两个子目录。
-
+```bash
 .
 ├── gridinfo # 用于dwa模块的参数文件
 ├── images   # 测试图片
 ├── wgt     # 用于拼接的权重文件
 └── videos   # 测试视频
-
+```
 
 ## 4. 环境准备
 
@@ -68,7 +69,7 @@ insmod v4l2_os04e10.ko
 ```bash
 sudo -s
 mkdir -p /mnt/cfg/param
-cp data/param/cvi_sdr_bin /mnt/cfg/param
+cp data/cvi_sdr_bin /mnt/cfg/param
 ```
 
 

diff --git a/samples/dwa_blend_encode/pic/image.jpg b/samples/dwa_blend_encode/pic/image.jpg
diff --git a/samples/dwa_dpu_encode/README.md b/samples/dwa_dpu_encode/README.md
@@ -62,7 +62,7 @@ chmod -R +x scripts/
 ```bash
 sudo -s
 cd /mnt/system/ko/
-insmod v4l2_os04a10.ko
+insmod v4l2_os04a10_sync.ko
 ```
 
 （2）isp参数文件配置

diff --git a/samples/gdwa_blend_encode/README.md b/samples/gdwa_blend_encode/README.md
@@ -24,7 +24,7 @@
 本例程用于说明如何使用sophon-stream快速构建深度估计应用。
 
 本例程中，广角拼接算法的广角展开、广角拼接分别在两个element上进行运算，element内部可以开启多个线程，保证了一定的运行效率。下图是广角拼接应用的流程图：
-![gdwa_pipeline](pic/image.png)
+![gdwa_pipeline](pic/image.jpg)
 ## 2. 特性
 
 * 支持BM1688(SoC)
@@ -43,13 +43,13 @@ chmod -R +x scripts/
 ```
 
 脚本执行完毕后，会在当前目录下生成`data`目录，其中包含`models`和`videos`两个子目录。
-
+```bash
 .
 ├── gridinfo # 用于dwa模块的参数文件
 ├── images   # 测试图片
 ├── wgt     # 用于拼接的权重文件
 └── videos   # 测试视频
-
+```
 
 ## 4. 环境准备
 
@@ -61,14 +61,14 @@ chmod -R +x scripts/
 ```bash
 sudo -s
 cd /mnt/system/ko/
-insmod v4l2_os04a10.ko
+insmod v4l2_os04a10_sync.ko
 ```
 （2）isp参数文件配置
 
 ```bash
 sudo -s
 mkdir -p /mnt/cfg/param
-cp data/param/cvi_sdr_bin /mnt/cfg/param
+cp data/cvi_sdr_bin /mnt/cfg/param
 ```