diff --git a/ACL_PyTorch/built-in/audio/LSTM/ReadMe.md b/ACL_PyTorch/built-in/audio/LSTM/ReadMe.md index 2897da6e701e578373ccf76eac49d2485a1d31af..9391b24d9f6c962c8a1a6549f705076c428e8116 100644 --- a/ACL_PyTorch/built-in/audio/LSTM/ReadMe.md +++ b/ACL_PyTorch/built-in/audio/LSTM/ReadMe.md @@ -316,7 +316,7 @@ LSTM是一种特殊的RNN模型,与普通RNN相比,LSTM可以更好地解决 4. 性能验证。 - 可使用ais_infer推理工具的纯推理模式验证不同batch_size的om模型的性能,参考命令如下: + 可使用ais_bench推理工具的纯推理模式验证不同batch_size的om模型的性能,参考命令如下: ``` python3 -m ais_bench --model=${om_model_path} --loop=20 --batchsize=${batch_size} diff --git a/ACL_PyTorch/built-in/cv/CascadeRCNN-DCN/README.md b/ACL_PyTorch/built-in/cv/CascadeRCNN-DCN/README.md index 15c4efa6de48534856d4735cb253015f9f77bc3b..525547a64ee91d0ebd60c92b1468a854f457cc3e 100644 --- a/ACL_PyTorch/built-in/cv/CascadeRCNN-DCN/README.md +++ b/ACL_PyTorch/built-in/cv/CascadeRCNN-DCN/README.md @@ -302,7 +302,7 @@ 4. 性能验证。 - 可使用ais_infer推理工具的纯推理模式验证不同batch_size的om模型的性能,参考命令如下: + 可使用ais_bench推理工具的纯推理模式验证不同batch_size的om模型的性能,参考命令如下: ``` python -m ais_bench --model ./cascade.om --loop 100 --batchsize 1 diff --git a/ACL_PyTorch/built-in/cv/SE_ResNet50_Pytorch_Infer/README.md b/ACL_PyTorch/built-in/cv/SE_ResNet50_Pytorch_Infer/README.md index 522873d50a8deb5f6673502ea95de84fa708d878..f8a2c5626c306d3ad7491a932af345b1708cc865 100644 --- a/ACL_PyTorch/built-in/cv/SE_ResNet50_Pytorch_Infer/README.md +++ b/ACL_PyTorch/built-in/cv/SE_ResNet50_Pytorch_Infer/README.md @@ -199,8 +199,6 @@ 推理后的输出默认在当前目录result下。 - >**说明:** - >执行ais-infer工具请选择与运行环境架构相同的命令。参数详情请参见。 3. 精度验证。 @@ -214,7 +212,7 @@ 执行后模型精度结果保存在./accuracy_result.json文件中 4. 性能验证。 - 可使用ais_infer推理工具的纯推理模式验证不同batch_size的om模型的性能,参考命令如下: + 可使用ais_bench推理工具的纯推理模式验证不同batch_size的om模型的性能,参考命令如下: ``` python3 -m ais_bench --model=${om_model_path} --loop=20 --batchsize=${batch_size} diff --git a/ACL_PyTorch/built-in/cv/SFA3D_for_Pytorch/README.md b/ACL_PyTorch/built-in/cv/SFA3D_for_Pytorch/README.md index 87252375cee0933473590c6c0f1552cab424fa01..0bcae36ddaf430b876717489bd3fad6cfc4d47d7 100644 --- a/ACL_PyTorch/built-in/cv/SFA3D_for_Pytorch/README.md +++ b/ACL_PyTorch/built-in/cv/SFA3D_for_Pytorch/README.md @@ -235,11 +235,11 @@ SFA3D(Super Fast and Accurate 3D Object Detection based on 3D LiDAR Point Clou 2. 执行推理。 ``` - mkdir ais_infer_result + mkdir ais_bench_result python3 -m ais_bench --model SFA3D_bs${n}.om --input ${input_data_save_path} --batchsize=${n} - --output ais_infer_result + --output ais_bench_result ``` 参数说明:\ @@ -248,7 +248,7 @@ SFA3D(Super Fast and Accurate 3D Object Detection based on 3D LiDAR Point Clou --input:模型输入,支持bin文件和目录,此例为数据文件夹路径。\ --output:推理结果输出路径。 - 推理后样本的输出在当前目录的ais_infer_result文件夹下,默认会建立日期+时间的子文件夹保存输出结果。 + 推理后样本的输出在当前目录的ais_bench_result文件夹下,默认会建立日期+时间的子文件夹保存输出结果。 3. 精度验证。 @@ -269,7 +269,7 @@ SFA3D(Super Fast and Accurate 3D Object Detection based on 3D LiDAR Point Clou 参数说明:\ --dataset_dir:KITTI数据集路径,默认路径为 "./SFA3D/dataset/kitti/" 目录。\ - --result_path:推理结果路径,默认为ais_infer_result目录下文件,例如 "./ais_infer_result/dumpdata_outputs/ "。 + --result_path:推理结果路径,默认为ais_bench_result目录下文件,例如 "./ais_bench_result/dumpdata_outputs/ "。 2. 开源模型loss统计。 diff --git a/ACL_PyTorch/contrib/audio/FastPitch/README.md b/ACL_PyTorch/contrib/audio/FastPitch/README.md index a49f4e196ae5173ee52d8d6e3b9e68366fa80d29..06c14f22a8859003b3e86fa5b0fff908cdbd48b4 100644 --- a/ACL_PyTorch/contrib/audio/FastPitch/README.md +++ b/ACL_PyTorch/contrib/audio/FastPitch/README.md @@ -228,7 +228,7 @@ Fastpitch模型由双向 Transformer 主干(也称为 Transformer 编码器) 5. 性能验证。 - 可使用ais_infer推理工具的纯推理模式验证不同batch_size的om模型的性能,参考命令如下: + 可使用ais_bench推理工具的纯推理模式验证不同batch_size的om模型的性能,参考命令如下: ``` python3 -m ais_bench --model=${om_model_path} --loop=20 --batchsize=${batch_size} diff --git a/ACL_PyTorch/contrib/cv/classfication/Swin-Transformer_tiny/README.md b/ACL_PyTorch/contrib/cv/classfication/Swin-Transformer_tiny/README.md index 73af06e21f37a04a0af2ec6ac454c5a9dc507932..a013f951df93759c224190f80e55c521cb521568 100644 --- a/ACL_PyTorch/contrib/cv/classfication/Swin-Transformer_tiny/README.md +++ b/ACL_PyTorch/contrib/cv/classfication/Swin-Transformer_tiny/README.md @@ -230,7 +230,7 @@ Swin-Transformer是针对于图片处理设计的基于Transformer架构的神 ``` python3 swin_postprocess.py --input_dir=outputs/bs16/ --label_path=val_label.txt --save_path=./result_bs16.json ``` - 注:--input_dir指定的路径不是固定,具体路径为ais-infer工具推理命令中'--output/--output_dirname'指定目录下的生成推理结果所在路径 + 注:--input_dir指定的路径不是固定,具体路径为ais_bench工具推理命令中'--output/--output_dirname'指定目录下的生成推理结果所在路径 --input_dir:生成推理结果所在路径 diff --git a/ACL_PyTorch/contrib/cv/detection/CTPN/README.md b/ACL_PyTorch/contrib/cv/detection/CTPN/README.md index d1a5c025fd4e57491a3bac21b5e90be65c1228d5..5ad1b4413bf8889d9bf7e26da2cf19f70f475b3e 100644 --- a/ACL_PyTorch/contrib/cv/detection/CTPN/README.md +++ b/ACL_PyTorch/contrib/cv/detection/CTPN/README.md @@ -247,10 +247,10 @@ CTPN是一种文字检测算法,它结合了CNN与LSTM深度网络,能有效 请访问[ais_bench推理工具](https://gitee.com/ascend/tools/tree/master/ais-bench_workload/tool/ais_bench)代码仓,根据readme文档进行工具安装。 - 2. 执行推理(${ais_infer_path}请根据实际的推理工具路径填写)。 + 2. 执行推理(${ais_bench_path}请根据实际的推理工具路径填写)。 ``` - python3 task_process.py --interpreter="python3 ${ais_infer_path}/ais_infer.py" --om_path=./ctpn_bs1.om --src_dir=./pre_bin/images_bin --res_dir=./result --batch_size=1 --device=0 + python3 task_process.py --interpreter="python3 ${ais_bench_path}/ais_infer.py" --om_path=./ctpn_bs1.om --src_dir=./pre_bin/images_bin --res_dir=./result --batch_size=1 --device=0 ``` - 参数说明: - --interpreter:推理工具。 diff --git a/ACL_PyTorch/contrib/cv/detection/Detr/README.md b/ACL_PyTorch/contrib/cv/detection/Detr/README.md index 13be0cd4d41cfe49b5745e36238dd36ebcad1946..0be42d0f9d4e54dac63e0f88bd753686901ed653 100755 --- a/ACL_PyTorch/contrib/cv/detection/Detr/README.md +++ b/ACL_PyTorch/contrib/cv/detection/Detr/README.md @@ -232,7 +232,7 @@ DETR是将目标检测视为一个集合预测问题(集合其实和anchors的 - --ais_path:ais_bench推理工具推理文件路径 - --img_path:前处理的图片文件路径 - --mask_path:前处理的mask文件路径 - - --out_put:ais_infer推理数据输出路径 + - --out_put:ais_bench推理数据输出路径 - --result:推理数据最终汇总路径 - --batch_size:batch大小,可选1或4 diff --git a/ACL_PyTorch/contrib/cv/detection/FCENet/readme.md b/ACL_PyTorch/contrib/cv/detection/FCENet/readme.md index 217d9ed57e6f96dd67918363e7e4be4053413288..45b2fad134a80ff97a41c4f16d4371f2a565243b 100644 --- a/ACL_PyTorch/contrib/cv/detection/FCENet/readme.md +++ b/ACL_PyTorch/contrib/cv/detection/FCENet/readme.md @@ -169,7 +169,7 @@ FCENet,使用傅里叶变换来得到文本的包围框,该方法在弯曲 ## 推理验证 1. 对数据集推理 - 该离线模型使用ais_infer作为推理工具,请参考[**安装文档**](https://gitee.com/ascend/tools/tree/master/ais-bench_workload/tool/ais_bench#%E4%B8%80%E9%94%AE%E5%AE%89%E8%A3%85)安装推理后端包aclruntime与推理前端包ais_bench。完成安装后,执行以下命令预处理后的数据进行推理。 + 安装ais_bench推理工具。请访问[ais_bench推理工具](https://gitee.com/ascend/tools/tree/master/ais-bench_workload/tool/ais_bench)代码仓,根据readme文档进行工具安装。 ```bash python3 -m ais_bench \ --model ./fcenet_bs${batch_size} \ diff --git a/ACL_PyTorch/contrib/cv/detection/NAS_FPN/README.md b/ACL_PyTorch/contrib/cv/detection/NAS_FPN/README.md index a7753d0fa6d2181a1d07fe0d0fde62c3e29fb578..a13223627e26518127f71107e70478399d6f4c39 100644 --- a/ACL_PyTorch/contrib/cv/detection/NAS_FPN/README.md +++ b/ACL_PyTorch/contrib/cv/detection/NAS_FPN/README.md @@ -268,7 +268,7 @@ python3 mmdetection_coco_postprocess.py --bin_data_path=../result/2022_xx_xx-xx_ --img_path:推理数据集。 ---is_ais_infer: 是否使用的是ais_infer进行推理。 +--is_ais_infer: 是否使用的是ais_bench进行推理。 评测结果的mAP值需要使用官方的pycocotools工具,首先将后处理输出的txt文件转化为coco数据集评测精度的标准json格式。 diff --git a/ACL_PyTorch/contrib/cv/detection/YOLOV3/README.md b/ACL_PyTorch/contrib/cv/detection/YOLOV3/README.md index d2de3baca59a06358673ce20bb30375f7b7c6cfc..60d781b4ac783dd239c79dc52f33d5fc5db77395 100755 --- a/ACL_PyTorch/contrib/cv/detection/YOLOV3/README.md +++ b/ACL_PyTorch/contrib/cv/detection/YOLOV3/README.md @@ -264,12 +264,12 @@ YOLOv3是一种端到端的one-stage目标检测模型。相比与YOLOv2,YOLOv b. 执行推理。 - mkdir ais_infer_result + mkdir ais_bench_result source /usr/local/Ascend/ascend-toolkit/set_env.sh python3 -m ais_bench --model yolov3_bsn.om --input yolov3_bin --batchsize=n - --output ais_infer_result + --output ais_bench_result 推理后的输出默认在当前目录result下。 @@ -277,11 +277,11 @@ YOLOv3是一种端到端的one-stage目标检测模型。相比与YOLOv2,YOLOv c. 模型后处理。 解析输出特征图。 - 解析ais_infer输出文件,经过阈值过滤,nms,坐标转换等输出坐标信息和类别信息txt文件。 + 解析ais_bench输出文件,经过阈值过滤,nms,坐标转换等输出坐标信息和类别信息txt文件。 python3.7 bin_to_predict_yolo_pytorch.py - --bin_data_path ais_infer_result/${ais_infer输出的结果}/ + --bin_data_path ais_bench_result/${ais_bench输出的结果}/ --det_results_path detection-results/ --origin_jpg_path val2014/ --coco_class_names coco2014.names diff --git a/ACL_PyTorch/contrib/cv/detection/YOLOX/readme.md b/ACL_PyTorch/contrib/cv/detection/YOLOX/readme.md index b2687a61ad14367a3366d460fb5b0e6b0aa6bc07..a5e5a4c74a9bbfad1bb69606968379ed336f96cf 100644 --- a/ACL_PyTorch/contrib/cv/detection/YOLOX/readme.md +++ b/ACL_PyTorch/contrib/cv/detection/YOLOX/readme.md @@ -224,7 +224,7 @@ YOLOX是基于往年对YOLO系列众多改进而产生的目标检测模型, - 参数说明: - --dataroot:数据集路径 - - --dump_dir:`ais-infer`推理结果文件目录 + - --dump_dir:`ais_bench`推理结果文件目录 # 模型推理性能&精度 diff --git a/ACL_PyTorch/contrib/cv/detection/ch_ppocr_server_v2.0_det/README.md b/ACL_PyTorch/contrib/cv/detection/ch_ppocr_server_v2.0_det/README.md index f7f0e10c52fead661de9631f8163e89b289007f4..45c5001201958ab739a442e14414b09f20c16fab 100644 --- a/ACL_PyTorch/contrib/cv/detection/ch_ppocr_server_v2.0_det/README.md +++ b/ACL_PyTorch/contrib/cv/detection/ch_ppocr_server_v2.0_det/README.md @@ -217,13 +217,13 @@ ch_PP-OCRv2_det是基于PP-OCRv2的中文文本检测模型,PP-OCRv2在PP-OCR b. 执行推理。 ``` python3 ch_server_det_ais_infer.py \ - --ais_infer=${path_to_ais-infer}/ais_infer.py \ + --ais_infer=${path_to_ais_bench}/ais_infer.py \ --model=./ch_ppocr_server_det_bs${batchsize}.om \ --inputs=./pre_data \ --batchsize=${batchsize} ``` - `${path_to_ais-infer}`为ais_infer.py脚本的存放路径。`${batchsize}`表示不同batch的om模型。 + `${path_to_ais_bench}`为ais_infer.py脚本的存放路径。`${batchsize}`表示不同batch的om模型。 c. 精度验证。 diff --git a/ACL_PyTorch/contrib/cv/detection/en_PP-OCRv3_det/README.md b/ACL_PyTorch/contrib/cv/detection/en_PP-OCRv3_det/README.md index 65e21a4d7821048189f18b9ce57d5d4e4ebc9fa9..a5ddcbf3ed27592946aa9bfbbf10b4f3e4679b15 100644 --- a/ACL_PyTorch/contrib/cv/detection/en_PP-OCRv3_det/README.md +++ b/ACL_PyTorch/contrib/cv/detection/en_PP-OCRv3_det/README.md @@ -230,7 +230,7 @@ en_PP-OCRv3_det是基于[[PP-OCRv3](https://github.com/PaddlePaddle/PaddleOCR/bl ``` python en_PP-OCRv3_det_ais_infer.py \ - --ais_infer=${path_to_ais-infer}/ais_infer.py \ + --ais_infer=${path_to_ais_bench}/ais_infer.py \ --model=./en_PP-OCRv3_det_bs${batchsize}.om \ --inputs=./pre_data \ --batchsize=${batchsize} @@ -242,7 +242,7 @@ en_PP-OCRv3_det是基于[[PP-OCRv3](https://github.com/PaddlePaddle/PaddleOCR/bl - --inputs:输入数据集路径。 - --batchsize:om模型的batchsize。 - `${path_to_ais-infer}`为ais_infer.py脚本的存放路径。`${batchsize}`表示不同batch的om模型。。 + `${path_to_ais_bench}`为ais_infer.py脚本的存放路径。`${batchsize}`表示不同batch的om模型。。 推理完成后结果保存在`en_PP-OCRv3_det/results_bs${batchsize}`目录下。 diff --git a/ACL_PyTorch/contrib/cv/detection/yolor/README.md b/ACL_PyTorch/contrib/cv/detection/yolor/README.md index 067fd1ec194fccab1abc13889f04fa108d1cbc1e..90a08c6c8ea8b3193c16ac876be200995a794b4b 100644 --- a/ACL_PyTorch/contrib/cv/detection/yolor/README.md +++ b/ACL_PyTorch/contrib/cv/detection/yolor/README.md @@ -245,7 +245,7 @@ b. 执行推理。 c. 精度验证。 -调用yolor_postprocess.py,可以获得Accuracy数据。修改yolor_postprocess.py第109行output_path为ais_infer推理的output路径。 +调用yolor_postprocess.py,可以获得Accuracy数据。修改yolor_postprocess.py第109行output_path为ais_bench推理的output路径。 ``` python3 yolor_postprocess.py --data ./coco.yaml --img 1344 --batch 1 --conf 0.001 --iou 0.65 --npu 0 --name yolor_p6_val --names ./yolor/data/coco.names diff --git a/ACL_PyTorch/contrib/cv/detection/yolor/yolor_postprocess.py b/ACL_PyTorch/contrib/cv/detection/yolor/yolor_postprocess.py index 92e495ad9140990d5498fb5322732e3e10544e17..f86d7a522ad130c30b6f198c84f45c9f333dc830 100644 --- a/ACL_PyTorch/contrib/cv/detection/yolor/yolor_postprocess.py +++ b/ACL_PyTorch/contrib/cv/detection/yolor/yolor_postprocess.py @@ -106,7 +106,7 @@ def test(data, loss = torch.zeros(3, device=device) jdict, stats, ap, ap_class, wandb_images = [], [], [], [], [] # output_path = './result/dumpOutput_device0' - output_path = './tools/ais-bench_workload/tool/ais_infer/2022_09_20-09_55_08' + output_path = './tools/ais-bench_workload/tool/ais_bench/2022_09_20-09_55_08' pbar = tqdm(dataloader) # ort_session = ort.InferenceSession('yolor_bs1simp.onnx') diff --git a/ACL_PyTorch/contrib/cv/gan/GAN/README.md b/ACL_PyTorch/contrib/cv/gan/GAN/README.md index 898049472b38589f7e143dbb21bf00db6dfd0951..6d74ae2b1d5921436ebf8890ea7a574d2e07a6a7 100644 --- a/ACL_PyTorch/contrib/cv/gan/GAN/README.md +++ b/ACL_PyTorch/contrib/cv/gan/GAN/README.md @@ -235,7 +235,7 @@ d. 性能验证。 - ais_infer纯推理验证不同batch_size的om模型的性能,参考命令如下: + ais_bench纯推理验证不同batch_size的om模型的性能,参考命令如下: ``` python3.7 -m ais_bench --model=${om_model_path} --loop=100 --batchsize=${batch_size} diff --git a/ACL_PyTorch/contrib/cv/image_registration/superpoint/README.md b/ACL_PyTorch/contrib/cv/image_registration/superpoint/README.md index 6e73a534c83f07f0aa80aa0d8979ff7998355652..02535955deab3cf7f6821d8b04c08efc32bc4b9a 100644 --- a/ACL_PyTorch/contrib/cv/image_registration/superpoint/README.md +++ b/ACL_PyTorch/contrib/cv/image_registration/superpoint/README.md @@ -230,8 +230,7 @@ python -m ais_bench --model=${om_model_path} --loop=20 --batchsize=${batch_size} ``` | 参数 | 说明 - | -------- |---------------------------| - | ais_infer_path | ais_infer文件路径 | + | -------- |---------------------------| | om_model_path | 模型文件保存的位置 | |batchsize | batchsize大小 | diff --git a/ACL_PyTorch/contrib/cv/pose_estimation/PoseC3D/README.md b/ACL_PyTorch/contrib/cv/pose_estimation/PoseC3D/README.md index 18e44f080aef4a81d32ea36dd04f46815af8cda1..161048dacc074419f21d6ea37bc0bbc55f9a8f22 100644 --- a/ACL_PyTorch/contrib/cv/pose_estimation/PoseC3D/README.md +++ b/ACL_PyTorch/contrib/cv/pose_estimation/PoseC3D/README.md @@ -177,7 +177,7 @@ PoseC3D 是一种基于 3D-CNN 的骨骼行为识别框架,同时具备良好 ## 推理验证 1. 对数据集推理 - 该离线模型使用ais_infer作为推理工具,请参考[**安装文档**](https://gitee.com/ascend/tools/tree/master/ais-bench_workload/tool/ais_bench#%E4%B8%80%E9%94%AE%E5%AE%89%E8%A3%85)安装推理后端包aclruntime与推理前端包ais_bench。完成安装后,执行以下命令预处理后的数据进行推理。 + 安装ais_bench推理工具。请访问[ais_bench推理工具](https://gitee.com/ascend/tools/tree/master/ais-bench_workload/tool/ais_bench)代码仓,根据readme文档进行工具安装。 ```bash python -m ais_bench \ --model ./posec3d_bs${bs}.om \ diff --git a/ACL_PyTorch/contrib/cv/quality_enhancement/ADNet/README.md b/ACL_PyTorch/contrib/cv/quality_enhancement/ADNet/README.md index 42826f00233b951f224c4364ecad29547cdce476..560de88bcc70a4ac069deb8f86d877705f024d33 100644 --- a/ACL_PyTorch/contrib/cv/quality_enhancement/ADNet/README.md +++ b/ACL_PyTorch/contrib/cv/quality_enhancement/ADNet/README.md @@ -257,7 +257,7 @@ d. 性能验证。 - 可使用ais_infer推理工具的纯推理模式验证不同batch_size的om模型的性能,参考命令如下: + 可使用ais_bench推理工具的纯推理模式验证不同batch_size的om模型的性能,参考命令如下: ```python python3 -m ais_bench --model=${om_model_path} --loop=20 --batchsize=${batch_size} --device=${device_id} --outfmt=BIN diff --git a/ACL_PyTorch/contrib/cv/segmentation/DeeplabV3/README.md b/ACL_PyTorch/contrib/cv/segmentation/DeeplabV3/README.md index 3a3dfada91240c0294b0352faa8aba44b8ae9e50..f69bfc949d37ca6419b950c371cf6d7692eba417 100644 --- a/ACL_PyTorch/contrib/cv/segmentation/DeeplabV3/README.md +++ b/ACL_PyTorch/contrib/cv/segmentation/DeeplabV3/README.md @@ -239,7 +239,7 @@ DeeplabV3是一个经典的图像语义分割网络,在v1和v2版本基础上 ``` - 参数说明: - - --output_path:ais_infer生成推理结果所在路径。 + - --output_path:ais_bench生成推理结果所在路径。 - --gt_path:标签数据路径。 - --result_path:为生成结果文件 diff --git a/ACL_PyTorch/contrib/cv/segmentation/ErfNet/README.md b/ACL_PyTorch/contrib/cv/segmentation/ErfNet/README.md index 6f3bc723b32e1f5b878bf7b9873c757072a9b6cc..67ab051a6b2d2dc4cbac69d22ca25687c7fe3615 100644 --- a/ACL_PyTorch/contrib/cv/segmentation/ErfNet/README.md +++ b/ACL_PyTorch/contrib/cv/segmentation/ErfNet/README.md @@ -193,7 +193,7 @@ python3 -m ais_bench --model ${user_path}/ErfNet/ErfNet_bs1.om --input=${user_pa python ErfNet_postprocess.py ${user_path}/output/2022_07_15-14_16_46/sumary.json ${user_path}/ErfNet/gt_label/ ``` -“${user_path}/output/2022_07_15-14_16_46/sumary.json”:ais_infer推理结果汇总数据保存路径。 +“${user_path}/output/2022_07_15-14_16_46/sumary.json”:ais_bench推理结果汇总数据保存路径。 ${user_path}/ErfNet/gt_label/:合并后的验证集路径。 diff --git a/ACL_PyTorch/contrib/cv/segmentation/OCRNet/README.md b/ACL_PyTorch/contrib/cv/segmentation/OCRNet/README.md index 0c5159d55a47376ef33a84cfe08eca076f75e14f..b9b844cb681da712234965126676bae10225def1 100644 --- a/ACL_PyTorch/contrib/cv/segmentation/OCRNet/README.md +++ b/ACL_PyTorch/contrib/cv/segmentation/OCRNet/README.md @@ -205,7 +205,8 @@ ## 推理验证 -1. 该离线模型使用ais_infer作为推理工具,请参考[**安装文档**](https://gitee.com/ascend/tools/tree/master/ais-bench_workload/tool/ais_bench#%E4%B8%80%E9%94%AE%E5%AE%89%E8%A3%85)安装推理后端包aclruntime与推理前端包ais_bench。完成安装后,执行以下命令预处理后的数据进行推理。 +1. 安装ais_bench推理工具。 + 请访问[ais_bench推理工具](https://gitee.com/ascend/tools/tree/master/ais-bench_workload/tool/ais_bench)代码仓,根据readme文档进行工具安装。 ```bash python3 -m ais_bench --model ./ocrnet_optimize_bs${batch_size}.om \ diff --git a/ACL_PyTorch/contrib/cv/segmentation/Transformer-SSL/README.md b/ACL_PyTorch/contrib/cv/segmentation/Transformer-SSL/README.md index 4c93845567a05097927ae86654a9ac88bee21add..5183f006f90365dc95fde74b0e3faa87e6e9cced 100755 --- a/ACL_PyTorch/contrib/cv/segmentation/Transformer-SSL/README.md +++ b/ACL_PyTorch/contrib/cv/segmentation/Transformer-SSL/README.md @@ -239,7 +239,7 @@ Transformer-SSL使用不同的IOU阈值,训练多个级联的检测器。它 - 参数说明: - --ann_file_path:为标签信息文件。 - - --bin_file_path:为ais_infer推理结果存放路径。 + - --bin_file_path:为ais_bench推理结果存放路径。 # 模型推理性能&精度 该模型只支持bs1推理 diff --git a/ACL_PyTorch/contrib/cv/segmentation/Ultra-Fast-Lane-Detection/README.md b/ACL_PyTorch/contrib/cv/segmentation/Ultra-Fast-Lane-Detection/README.md index cdf0126c64cac5b9b7c1ea83e828c2dbe69df167..b03a2413d2e4c5cd1e970f1b040b38e7a2a6751c 100644 --- a/ACL_PyTorch/contrib/cv/segmentation/Ultra-Fast-Lane-Detection/README.md +++ b/ACL_PyTorch/contrib/cv/segmentation/Ultra-Fast-Lane-Detection/README.md @@ -208,7 +208,7 @@ 2. 离线推理 - 使用ais_infer工具将预处理后的数据传入模型并执行推理: + 使用ais_bench工具将预处理后的数据传入模型并执行推理: ```shell # 设置环境变量 source /usr/local/Ascend/ascend-toolkit/set_env.sh diff --git a/ACL_PyTorch/contrib/cv/segmentation/YOLACT/README.md b/ACL_PyTorch/contrib/cv/segmentation/YOLACT/README.md index bce6ca4a2e513a758abfff95e12445f9fbeed569..ab2b6c2d2490a77197a2d91397eb068dc78c79b6 100644 --- a/ACL_PyTorch/contrib/cv/segmentation/YOLACT/README.md +++ b/ACL_PyTorch/contrib/cv/segmentation/YOLACT/README.md @@ -214,7 +214,7 @@ YOLACT是2019年发表在ICCV上面的一个实时实例分割的模型,它主 请访问[ais_bench推理工具](https://gitee.com/ascend/tools/tree/master/ais-bench_workload/tool/ais_bench)代码仓,根据readme文档进行工具安装。 - 2. 执行推理(${ais_infer_path}请根据实际的推理工具路径填写)。 + 2. 执行推理。 ``` mkdir result diff --git a/ACL_PyTorch/contrib/cv/super_resolution/RCAN/README.md b/ACL_PyTorch/contrib/cv/super_resolution/RCAN/README.md index 0b39c810923301141405f62fc470dc9194b12abf..d7393073fcdc324df1e4c03df89a161de95e3b13 100644 --- a/ACL_PyTorch/contrib/cv/super_resolution/RCAN/README.md +++ b/ACL_PyTorch/contrib/cv/super_resolution/RCAN/README.md @@ -169,7 +169,7 @@ RCAN设计了一个残差中的残差(RIR)结构来构造深层网络,每 ## 推理验证 1. 对数据集推理 - 该离线模型使用ais_infer作为推理工具,请参考[**安装文档**](https://gitee.com/ascend/tools/tree/master/ais-bench_workload/tool/ais_bench#%E4%B8%80%E9%94%AE%E5%AE%89%E8%A3%85)安装推理后端包aclruntime与推理前端包ais_bench。完成安装后,执行以下命令预处理后的数据进行推理。 + 安装ais_bench推理工具。请访问[ais_bench推理工具](https://gitee.com/ascend/tools/tree/master/ais-bench_workload/tool/ais_bench)代码仓,根据readme文档进行工具安装。 ```bash python -m ais_bench \ --model rcan_bs${bs}.om \ diff --git a/ACL_PyTorch/contrib/cv/video_understanding/SlowFast/README.md b/ACL_PyTorch/contrib/cv/video_understanding/SlowFast/README.md index 79380e8f133444636d6ed372684cdb6f2e0d2c56..b4b0f48e21d60c2f2cf399e406142b6521757e34 100644 --- a/ACL_PyTorch/contrib/cv/video_understanding/SlowFast/README.md +++ b/ACL_PyTorch/contrib/cv/video_understanding/SlowFast/README.md @@ -193,7 +193,7 @@ SlowFast 是用于视频理解的双流框架的卷积神经网络,该网络 ## 推理验证 1. 对数据集推理 - 该离线模型使用ais_infer作为推理工具,请参考[**安装文档**](https://gitee.com/ascend/tools/tree/master/ais-bench_workload/tool/ais_bench#%E4%B8%80%E9%94%AE%E5%AE%89%E8%A3%85)安装推理后端包aclruntime与推理前端包ais_bench。完成安装后,执行以下命令预处理后的数据进行推理。 + 安装ais_bench推理工具。请访问[ais_bench推理工具](https://gitee.com/ascend/tools/tree/master/ais-bench_workload/tool/ais_bench)代码仓,根据readme文档进行工具安装。 ```bash python -m ais_bench \ --model slowfast_bs${bs}.om \ diff --git a/ACL_PyTorch/contrib/cv/video_understanding/X3D/README.md b/ACL_PyTorch/contrib/cv/video_understanding/X3D/README.md index 93799cc9a58ef452354fdbe62ecedbbf8134afe5..fd03b750ac6fecccd92b2b8d5b2ffea1760382e1 100644 --- a/ACL_PyTorch/contrib/cv/video_understanding/X3D/README.md +++ b/ACL_PyTorch/contrib/cv/video_understanding/X3D/README.md @@ -180,7 +180,7 @@ X3D,这是一个高效的视频网络系列,可沿多个网络轴在空间 ## 推理验证 1. 对数据集推理 - 该离线模型使用ais_infer作为推理工具,请参考[**安装文档**](https://gitee.com/ascend/tools/tree/master/ais-bench_workload/tool/ais_bench#%E4%B8%80%E9%94%AE%E5%AE%89%E8%A3%85)安装推理后端包aclruntime与推理前端包ais_bench。完成安装后,执行以下命令预处理后的数据进行推理。 + 安装ais_bench推理工具。请访问[ais_bench推理工具](https://gitee.com/ascend/tools/tree/master/ais-bench_workload/tool/ais_bench)代码仓,根据readme文档进行工具安装。 ```bash python -m ais_bench \ --model x3d_s_bs${bs}.om \