peace195 / Sppnet
Spatial Pyramid Pooling on top of AlexNet using tensorflow. ***New updates for SPPnet in Pytorch**
Stars: ✭ 117
Programming Languages
python
139335 projects - #7 most used programming language
Projects that are alternatives of or similar to Sppnet
Deepface
Keras implementation of the renowned publication "DeepFace: Closing the Gap to Human-Level Performance in Face Verification" by Taigman et al. Pre-trained weights on VGGFace2 dataset.
Stars: ✭ 113 (-3.42%)
Mutual labels: cnn
Deep learning for computer vision with python
Deep Learning For Computer Vision With Python
Stars: ✭ 106 (-9.4%)
Mutual labels: cnn
Mnist Classification
Pytorch、Scikit-learn实现多种分类方法,包括逻辑回归(Logistic Regression)、多层感知机(MLP)、支持向量机(SVM)、K近邻(KNN)、CNN、RNN,极简代码适合新手小白入门,附英文实验报告(ACM模板)
Stars: ✭ 109 (-6.84%)
Mutual labels: cnn
Mobilenet Yolo
MobileNetV2-YoloV3-Nano: 0.5BFlops 3MB HUAWEI P40: 6ms/img, YoloFace-500k:0.1Bflops 420KB🔥🔥🔥
Stars: ✭ 1,566 (+1238.46%)
Mutual labels: cnn
Yolo V3 Iou
YOLO3 动漫人脸检测 (Based on keras and tensorflow) 2019-1-19
Stars: ✭ 116 (-0.85%)
Mutual labels: cnn
Deeplearning tutorials
The deeplearning algorithms implemented by tensorflow
Stars: ✭ 1,580 (+1250.43%)
Mutual labels: cnn
Deepgaze
Computer Vision library for human-computer interaction. It implements Head Pose and Gaze Direction Estimation Using Convolutional Neural Networks, Skin Detection through Backprojection, Motion Detection and Tracking, Saliency Map.
Stars: ✭ 1,552 (+1226.5%)
Mutual labels: cnn
Awesome Deep Learning Resources
Rough list of my favorite deep learning resources, useful for revisiting topics or for reference. I have got through all of the content listed there, carefully. - Guillaume Chevalier
Stars: ✭ 1,469 (+1155.56%)
Mutual labels: cnn
Self Driving Car
A End to End CNN Model which predicts the steering wheel angle based on the video/image
Stars: ✭ 106 (-9.4%)
Mutual labels: cnn
Id Card Detector
💳 Detecting the National Identification Cards with Deep Learning (Faster R-CNN)
Stars: ✭ 114 (-2.56%)
Mutual labels: cnn
Pwc Net pytorch
pytorch implementation of "PWC-Net: CNNs for Optical Flow Using Pyramid, Warping, and Cost Volume"
Stars: ✭ 111 (-5.13%)
Mutual labels: cnn
Aspect Extraction
Aspect extraction from product reviews - window-CNN+maxpool+CRF, BiLSTM+CRF, MLP+CRF
Stars: ✭ 117 (+0%)
Mutual labels: cnn
Reproducible Image Denoising State Of The Art
Collection of popular and reproducible image denoising works.
Stars: ✭ 1,776 (+1417.95%)
Mutual labels: cnn
Tf2
An Open Source Deep Learning Inference Engine Based on FPGA
Stars: ✭ 113 (-3.42%)
Mutual labels: cnn
Spatial Pyramid Pooling in Deep Convolutional Networks using tensorflow
New updates
Instead of sppnet, you can use this block of code in Pytorch to train a neural network with variable-sized inputs:
#With these lines of code below, we can memorize the gradient for later updates using pytorch because the
#loss.backward()function accumulates the gradient. After 64 steps, we call optimizer.step() for updating the parameters.
#https://discuss.pytorch.org/t/how-are-optimizer-step-and-loss-backward-related/7350
train_loader = torch.utils.data.DataLoader(dataset=train_dataset, batch_size=1, num_workers=8, shuffle=True)
test_loader = torch.utils.data.DataLoader(dataset=test_dataset, batch_size=1, num_workers=8, shuffle=False)
for i, (seqs, labels) in enumerate(train_loader):
...
loss = criterion(outputs, labels)
loss.backward()
if i % 64 == 0 or i == len(train_loader) - 1:
optimizer.step()
optimizer.zero_grad()
...
Descriptions
I implemented a Spatial Pyramid Pooling on top of AlexNet in tensorflow. Then I applied it to 102 Category Flower identification task. I implemented for identification task only. If you are interested in this project, I will continue to develop it in object detection task. Do not hesitate to contact me at [email protected]. :)
More information: https://peace195.github.io/spatial-pyramid-pooling/
Data
Requirements
- python 2.7
- tensorflow 1.2
- pretrained parameters of AlexNet in ImageNet dataset: bvlc_alexnet.npy
Running
$ python alexnet_spp.py
Result
82% accuracy rate (the state-of-the-art is 94%).
Author
Binh Do
Note that the project description data, including the texts, logos, images, and/or trademarks,
for each open source project belongs to its rightful owner.
If you wish to add or remove any projects, please contact us at [email protected].