mit-han-lab/temporal-shift-module issues and pull requests

#235 - view size is not compatible with input tensor's size and stride (at least one dimension spans across two contiguous subspaces).

Issue - State: open - Opened by anuj018 about 2 months ago

#234 - TypeError: 'IntImm' object cannot be interpreted as an integer

Issue - State: open - Opened by SlowMonk 7 months ago

#233 - How to train pretrain model with Kinetics dataset? About choosing parameters

Issue - State: open - Opened by Eurususu 10 months ago

#232 - Which parameter in the code corresponds to the channel shift ratio? Is it 'shift_div'?

Issue - State: open - Opened by Changwei-Ouyang about 1 year ago

#231 - TSM model checkpoint download links not working

Issue - State: closed - Opened by wrko over 1 year ago - 2 comments

#230 - questions about raspberrypi

Issue - State: open - Opened by lujuncong2000 almost 2 years ago

#229 - nll_loss_forward_reduce_cuda_kernel_2d: block: [0,0,0], thread: [0,0,0] Assertion `t >= 0 && t < n_classes` failed.

Issue - State: closed - Opened by cherylngsy almost 2 years ago

#228 - Confusion about the parameter.

Issue - State: closed - Opened by ForeverPs almost 2 years ago

#227 - On the precision of something-something-v1

Issue - State: open - Opened by Maojianzeng almost 2 years ago

#226 - Question about online demo

Issue - State: open - Opened by andrewwang0612 almost 2 years ago

#225 - How to extract my own dataset and where is the pre-trained model？

Issue - State: open - Opened by coaast123 about 2 years ago

#224 - the pretrained model link is broken. It shows as text but not downloading

Issue - State: closed - Opened by SHMCU about 2 years ago - 2 comments

#223 - How to set up the correct combination of shift_div and frame_count?

Issue - State: open - Opened by qwangku over 2 years ago - 1 comment

#222 - Can not download the pretrained models

Issue - State: closed - Opened by CZC-026 over 2 years ago - 3 comments

#221 - Trying to understand fold_div

Issue - State: closed - Opened by Lucab95 over 2 years ago

#220 - Rebuild DPU kernel

Issue - State: open - Opened by lloo099 over 2 years ago

#219 - Acc1 about training HMDB51 pretrained on Imagenet

Issue - State: closed - Opened by VanChilly over 2 years ago - 5 comments

#218 - Excuse me, the parameters controlling the time shift are false. Where can I call them

Issue - State: open - Opened by syfffffff over 2 years ago

#217 - nvidia

Issue - State: closed - Opened by xxxlive over 2 years ago

#216 - Unable to understand average_duration > 0 and record.num_frames > self.num_segments:

Issue - State: open - Opened by ZubairKhan001 almost 3 years ago - 1 comment

#215 - Question on online google maps

Issue - State: open - Opened by niranjanravilevelup almost 3 years ago - 1 comment

#214 - 模型输出类别固定

Issue - State: closed - Opened by dengfenglai321 almost 3 years ago - 2 comments

#213 - 训练ucf101时getitem函数得返回值跟dataloader的返回值不一致

Issue - State: closed - Opened by dengfenglai321 almost 3 years ago

#212 - why the network's input shape is 1 24 224 224

Issue - State: closed - Opened by dengfenglai321 almost 3 years ago - 1 comment

#211 - the pretrained of TSM MobileNetV2

Issue - State: open - Opened by dengfenglai321 almost 3 years ago

#210 - The testing speed on Jeston Nano

Issue - State: closed - Opened by Z2HENG almost 3 years ago - 5 comments

#209 - The link of pretrained model is broken, please check it.Thanks!

Issue - State: closed - Opened by yubin2020 almost 3 years ago - 4 comments

#208 - How to train 16 frames TSM on something-something v2

Issue - State: open - Opened by zyc00 about 3 years ago

#207 - I am getting "ld.so: object '/' from LD_PRELOAD cannot be preloaded" error after running the script till step 5

Issue - State: closed - Opened by gangadharsingh056 about 3 years ago

#206 - When should I empty the shift_buffer?

Issue - State: open - Opened by waduhekx about 3 years ago

#205 - module 'torchvision' has no attribute 'transforms'

Issue - State: closed - Opened by zubairahmed-ai about 3 years ago - 2 comments

#204 - what is the correct shape for TSM expected?

Issue - State: open - Opened by mbilalshaikh about 3 years ago - 1 comment

#203 - Link to checkpoints is broken

Issue - State: closed - Opened by Fritskee about 3 years ago

#202 - train jester demo model use "Dense Sample" or "Uniform Sampling"?

Issue - State: open - Opened by zhunge about 3 years ago

#201 - Could anyone share a copy of Jester dataset？

Issue - State: open - Opened by jianghaojun about 3 years ago - 1 comment

#200 - TSM model checkpoint download links are unavailable!

Issue - State: closed - Opened by jianghaojun over 3 years ago - 3 comments

#199 - What is the purpose of shift_buffer in the onlinedemo code?

Issue - State: open - Opened by Deepakgthomas over 3 years ago

#198 - RuntimeError: NCCL Error 2: unhandled system error

Issue - State: open - Opened by waduhekx over 3 years ago - 2 comments

#197 - could you share a copy of something something v2

Issue - State: closed - Opened by caopulan over 3 years ago - 4 comments

#196 - Missing mobilenet_v2_tsm_v2 file

Issue - State: closed - Opened by kdy1999 over 3 years ago - 1 comment

#195 - About the number of params!!!

Issue - State: open - Opened by Yi-dashi over 3 years ago - 1 comment

#194 - Why the model of "e45.pth" is used when testing on sth-sth v1?

Issue - State: open - Opened by withbrightmoon over 3 years ago

#193 - Are the results of the UCF101 and HMDB51 data sets in Table 1 in your TSM paper only the results of split 1?

Issue - State: open - Opened by geek12138 over 3 years ago

#192 - how to train the online model

Issue - State: open - Opened by lenka844 over 3 years ago - 4 comments

#191 - As for how to train my dataset

Issue - State: open - Opened by dy-gdut over 3 years ago - 6 comments

#190 - The offline train is same to online train or not ?

Issue - State: closed - Opened by qixi666666 over 3 years ago - 14 comments

#189 - for online version, onnx seems not support some OP in forward( ):

Issue - State: open - Opened by chongyangwang-song over 3 years ago - 26 comments

#188 - Cannot find expanded_conv_shift in tf_models at specific commit.

Issue - State: open - Opened by zshanwei over 3 years ago

#187 - about file 'mobilenetv2_jester_online.pth.tar'

Issue - State: open - Opened by tjinjin95 over 3 years ago - 5 comments

#186 - Update README.md

Pull Request - State: closed - Opened by lmxyy over 3 years ago

#184 - Update README.md

Pull Request - State: open - Opened by KhanalBijay over 3 years ago

#183 - Installing onnxruntime for Jetson devices

Issue - State: closed - Opened by KhanalBijay over 3 years ago

#182 - ucf101

Issue - State: open - Opened by checheche222 over 3 years ago

#181 - Using PyTorch and TensorRT, the results were quite different

Issue - State: open - Opened by momohuangsha over 3 years ago - 4 comments

#180 - what(): CHECK failed: generated_database_->Add(encoded_file_descriptor, size): Aborted (core dumped)

Issue - State: open - Opened by MyungHwanSung over 3 years ago

#179 - Predicting on a single video?

Issue - State: closed - Opened by balaganeshmohan over 3 years ago - 1 comment

#178 - Question about F.relu(x + layer_func(x_shift)) in bn_inception.py

Issue - State: open - Opened by DuL1nk over 3 years ago - 1 comment

#177 - How is your kinetics dataset structured?

Issue - State: open - Opened by benbwf almost 4 years ago - 2 comments

#176 - Details of finetuning UCF101 and HMDB51

Issue - State: open - Opened by icyzhang0923 almost 4 years ago - 1 comment

#175 - how to recognize mulit actions meanwhile in one video usingTSM

Issue - State: open - Opened by Wallace00V almost 4 years ago

#174 - Visualizing TSM model with GRAD-Cam

Issue - State: closed - Opened by Fritskee almost 4 years ago

#173 - google::protobuf::FatalException

Issue - State: open - Opened by Marco-KEA almost 4 years ago - 2 comments

#172 - the ssv1 acc of using test scripts is the results on val dataset. how to get the results on the official test dataset ?

Issue - State: open - Opened by gitxdf almost 4 years ago - 2 comments

#171 - online-demo run on colab

Issue - State: open - Opened by dandingol03 almost 4 years ago - 1 comment

#170 - Fixed error while training with mobilenet architecture

Pull Request - State: closed - Opened by Nauman007 almost 4 years ago

#169 - 请问有划分ucf101和hmdb51的代码吗

Issue - State: closed - Opened by Evigouse almost 4 years ago - 1 comment

#168 - how to train on ucf101 from scratch?

Issue - State: open - Opened by xiadingZ almost 4 years ago - 2 comments

#166 - How to speed up for online model on Jetson Nano(only 0.7vid/s)?

Issue - State: open - Opened by Amazingren almost 4 years ago - 10 comments

#165 - Is the number of frames used for training is differ from the number of frames used for testing?

Issue - State: open - Opened by liming-ai almost 4 years ago - 1 comment

#163 - Can you provide dataset of online_demo ?

Issue - State: open - Opened by Syorst almost 4 years ago - 1 comment

#162 - Online demo green screen

Issue - State: open - Opened by FireWizard312 almost 4 years ago - 1 comment

#160 - How to fine-tune with 'TSM_kinetics_RGB_mobilenetv2_shift8_blockres_avg_segment8_e100_dense.pth'

Issue - State: open - Opened by soolone about 4 years ago - 3 comments

#157 - conv2d dimension issue in online demo script

Issue - State: open - Opened by dhruvapatil about 4 years ago - 4 comments

#156 - No matching distribution found for onnxoptimizer>=0.1.1 on Jetson nano

Issue - State: open - Opened by lyf524951805 about 4 years ago - 3 comments

#154 - Use RWF-2000 to train TSM

Issue - State: open - Opened by bujianyiwang about 4 years ago - 4 comments

#153 - Why the input shape is (NT,C,H,W)?

Issue - State: open - Opened by zzwei1 about 4 years ago - 1 comment

#148 - TypeError: expected Tensor as element 0 in argument 0, but got tuple

Issue - State: closed - Opened by jainal09 about 4 years ago - 5 comments

#142 - Gesture image of the online demo

Issue - State: closed - Opened by czqInNanjing about 4 years ago - 9 comments

#140 - Optical Flow Results for SomethingV1 dataset

Issue - State: open - Opened by boheumd about 4 years ago - 1 comment

#138 - strided_slice get empty slice at axis 1

Issue - State: open - Opened by faheng about 4 years ago - 4 comments

#137 - Temporal shifts at multiple blocks / video object detection

Issue - State: open - Opened by alphadadajuju about 4 years ago - 1 comment

#136 - where can I find the 'kinetics_val.csv', 'kinetics_train.csv'

Issue - State: open - Opened by icyzhang0923 about 4 years ago - 2 comments

#134 - TSM with lower batch size decreases performance

Issue - State: open - Opened by ysl208 about 4 years ago - 1 comment

#103 - Convert TSM pytorch model to onnx

Issue - State: closed - Opened by Usernamezhx over 4 years ago - 8 comments

#100 - Can you tell me the environment required to run the code？

Issue - State: closed - Opened by chenjiadong96 over 4 years ago - 1 comment

#99 - test failed when using RGBDiff modality

Issue - State: open - Opened by sunnymoon155 over 4 years ago - 1 comment

#98 - online demo error with onnx opset

Issue - State: closed - Opened by poincarelee over 4 years ago - 6 comments

GitHub / mit-han-lab/temporal-shift-module issues and pull requests