-
**Is your feature request related to a problem? Please describe.**
Facial recognition systems are vulnerable to spoofing attacks (e.g., photos, videos, masks), making them insecure for certain applic…
-
## 🚀 Feature
LRP-based explanation rules that work during or similar to backwards passes.
## Motivation
I'm a PhD student working in explainability for Video Rec. models. I much prefer PyTorch to…
-
The goal is to figure out:
1. The most commonly used latest, such as YOLO family models, models used for object detection/tracking.
2. The most commonly used preprocessing operations to train thes…
-
Hi,
I am using the UCF Sports Action dataset which is kind of a subset of UCF101.
And I followed the instruction in Readme file to prepare the dataset.
However after I trained 200 epoch, the accura…
-
Similar to #3, I would like to apply this code to networks with 3D convolutions. The math is exactly the same, of course, but one needs to collect outputs along three dimensions instead of just two.
…
-
1、3D卷积核的shape中,那个恒定为1的维度代表了什么呢?之前您使用tf.expand_dims将输入的维度增加了1,这两者有关吗?kernel_size中的那些参数是根据什么来进行排列的呢?
2、tf.contrib.layers.conv3d提供了默认参数activation_fn=tf.nn.relu, 为什么之后还要手动进行一次激活?
3、with语句中的最后一行scope.reu…
-
https://arxiv.org/abs/1604.02426
-
Post your questions here about: “Image Learning” & “Audio and Video Learning”, Thinking with Deep Learning, Chapters 13 & 14.
-
Hi,
i am attempting to build a equivariant Variational Encoder-Decoder framework.
For this I am using R2Conv() and R3(Conv) layers in the encoder with trivial-representation input & output and …
-
Similar to benchmarks created for image classification, we would like to create a benchmark for video datasets.
A few tasks for tackling video datasets:
1. Identify ML training related to video …