From f3b392b3571e9cba0006a79c856003fc28c6a272 Mon Sep 17 00:00:00 2001 From: YouqingXiaozhua <843213558@qq.com> Date: Fri, 31 Mar 2023 21:34:04 +0800 Subject: [PATCH] add inference demo --- README.md | 4 + configs/_base_/datasets/RAF.py | 4 +- demo.ipynb | 195 +++++++++++++++++++++++++++++ mmcls/apis/inference.py | 18 +-- mmcls/models/vit/vit_siam_merge.py | 4 +- resources/demo.jpg | Bin 0 -> 6960 bytes 6 files changed, 213 insertions(+), 12 deletions(-) create mode 100644 demo.ipynb create mode 100644 resources/demo.jpg diff --git a/README.md b/README.md index 36d3c50..2dd5eae 100755 --- a/README.md +++ b/README.md @@ -7,6 +7,10 @@ APViT: Vision Transformer With Attentive Pooling for Robust Facial Expression Re APViT is a simple and efficient Transformer-based method for facial expression recognition (FER). It builds on the [TransFER](https://openaccess.thecvf.com/content/ICCV2021/html/Xue_TransFER_Learning_Relation-Aware_Facial_Expression_Representations_With_Transformers_ICCV_2021_paper.html), but introduces two attentive pooling (AP) modules that do not require any learnable parameters. These modules help the model focus on the most expressive features and ignore the less relevant ones. You can read more about our method in our [paper](https://arxiv.org/abs/2212.05463). +## Update + +- 2023-03-31: Added an [notebook demo](demo.ipynb) for inference. + ## Installation diff --git a/configs/_base_/datasets/RAF.py b/configs/_base_/datasets/RAF.py index 28ec461..21ae29a 100755 --- a/configs/_base_/datasets/RAF.py +++ b/configs/_base_/datasets/RAF.py @@ -34,8 +34,8 @@ dict(type='Resize', size=img_size), dict(type='Normalize', **img_norm_cfg), dict(type='ImageToTensor', keys=['img']), - dict(type='ToTensor', keys=['gt_label', ]), - dict(type='Collect', keys=['img', 'gt_label',]) + # dict(type='ToTensor', keys=['gt_label', ]), + dict(type='Collect', keys=['img', ]) ] base_path = 'data/RAF-DB/basic/' diff --git a/demo.ipynb b/demo.ipynb new file mode 100644 index 0000000..3f9c904 --- /dev/null +++ b/demo.ipynb @@ -0,0 +1,195 @@ +{ + "cells": [ + { + "attachments": {}, + "cell_type": "markdown", + "metadata": {}, + "source": [ + "# A simple demonstration to predict on an image" + ] + }, + { + "cell_type": "code", + "execution_count": 1, + "metadata": {}, + "outputs": [], + "source": [ + "import numpy as np\n", + "import mmcv\n", + "from mmcv.runner import load_checkpoint\n", + "\n", + "from mmcls.models.builder import build_classifier\n", + "from mmcls.datasets.raf import FER_CLASSES\n", + "from mmcls.datasets.pipelines import Compose" + ] + }, + { + "cell_type": "code", + "execution_count": 2, + "metadata": {}, + "outputs": [ + { + "name": "stdout", + "output_type": "stream", + "text": [ + "Unused kwargs: \n", + "{'img_size': 112, 'patch_size': 16}\n", + "load checkpoint from local path: weights/APViT_RAF-3eeecf7d.pth\n" + ] + } + ], + "source": [ + "cfg = mmcv.Config.fromfile(\"configs/apvit/RAF.py\")\n", + "cfg.model.pretrained = None\n", + "cfg.model.extractor.pretrained = None\n", + "cfg.model.vit.pretrained = None\n", + "\n", + "# build the model and load checkpoint\n", + "classifier = build_classifier(cfg.model)\n", + "load_checkpoint(classifier, \"weights/APViT_RAF-3eeecf7d.pth\", map_location='cpu')\n", + "classifier = classifier.to(\"cuda\")\n", + "classifier.eval()\n", + "\n", + "# define the preprocess for test\n", + "test_preprocess = Compose([\n", + " dict(type='Resize', size=112),\n", + " dict(type='Normalize',\n", + " mean=[123.675, 116.28, 103.53],\n", + " std=[58.395, 57.12, 57.375]),\n", + " dict(type='ImageToTensor', keys=['img']),\n", + " dict(type='Collect', keys=['img',])\n", + "])\n" + ] + }, + { + "cell_type": "code", + "execution_count": 3, + "metadata": {}, + "outputs": [ + { + "data": { + "text/plain": [ + "" + ] + }, + "execution_count": 3, + "metadata": {}, + "output_type": "execute_result" + }, + { + "data": { + "image/png": "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", + "text/plain": [ + "
" + ] + }, + "metadata": {}, + "output_type": "display_data" + } + ], + "source": [ + "img = mmcv.imread('resources/demo.jpg')\n", + "\n", + "import matplotlib.pyplot as plt\n", + "plt.imshow(img[:, :, ::-1])" + ] + }, + { + "cell_type": "code", + "execution_count": 4, + "metadata": {}, + "outputs": [ + { + "name": "stdout", + "output_type": "stream", + "text": [ + "Predict result: Happiness with confidance: 1.00\n" + ] + } + ], + "source": [ + "# preprocess the image\n", + "data = test_preprocess(dict(img=img))\n", + "data['img'] = data['img'][None, ...].cuda()\n", + "\n", + "# run the inference\n", + "out = classifier(**data, return_loss=False)\n", + "result_index = np.argmax(out[0])\n", + "\n", + "print(f'Predict result: {FER_CLASSES[result_index]} with confidance: {out[0][result_index]:.2f}')" + ] + }, + { + "attachments": {}, + "cell_type": "markdown", + "metadata": {}, + "source": [ + "## Alternatively, you can use high-level APIs from mmcls" + ] + }, + { + "cell_type": "code", + "execution_count": 5, + "metadata": {}, + "outputs": [ + { + "name": "stdout", + "output_type": "stream", + "text": [ + "Unused kwargs: \n", + "{'img_size': 112, 'patch_size': 16}\n", + "load checkpoint from local path: weights/APViT_RAF-3eeecf7d.pth\n" + ] + }, + { + "data": { + "text/plain": [ + "{'pred_label': 4, 'pred_score': 0.9999688863754272, 'pred_class': 'Happiness'}" + ] + }, + "execution_count": 5, + "metadata": {}, + "output_type": "execute_result" + } + ], + "source": [ + "from mmcls.apis.inference import init_model, inference_model\n", + "\n", + "model = init_model(\n", + " config='configs/apvit/RAF.py',\n", + " checkpoint='weights/APViT_RAF-3eeecf7d.pth'\n", + ")\n", + "\n", + "result = inference_model(model, img)\n", + "result" + ] + } + ], + "metadata": { + "kernelspec": { + "display_name": "mmdet", + "language": "python", + "name": "python3" + }, + "language_info": { + "codemirror_mode": { + "name": "ipython", + "version": 3 + }, + "file_extension": ".py", + "mimetype": "text/x-python", + "name": "python", + "nbconvert_exporter": "python", + "pygments_lexer": "ipython3", + "version": "3.9.15" + }, + "orig_nbformat": 4, + "vscode": { + "interpreter": { + "hash": "26983dda8997062e51c260bfdbd9127431e5c93a00e9b81f5a08036be419250a" + } + } + }, + "nbformat": 4, + "nbformat_minor": 2 +} diff --git a/mmcls/apis/inference.py b/mmcls/apis/inference.py index 61d161e..7e992aa 100755 --- a/mmcls/apis/inference.py +++ b/mmcls/apis/inference.py @@ -32,18 +32,20 @@ def init_model(config, checkpoint=None, device='cuda:0', options=None): if options is not None: config.merge_from_dict(options) config.model.pretrained = None + config.model.extractor.pretrained = None + config.model.vit.pretrained = None model = build_classifier(config.model) if checkpoint is not None: map_loc = 'cpu' if device == 'cpu' else None checkpoint = load_checkpoint(model, checkpoint, map_location=map_loc) - if 'CLASSES' in checkpoint['meta']: - model.CLASSES = checkpoint['meta']['CLASSES'] - else: - from mmcls.datasets import ImageNet - warnings.simplefilter('once') - warnings.warn('Class names are not saved in the checkpoint\'s ' - 'meta data, use imagenet by default.') - model.CLASSES = ImageNet.CLASSES + class_loaded = False + if 'meta' in checkpoint: + if 'CLASSES' in checkpoint['meta']: + model.CLASSES = checkpoint['meta']['CLASSES'] + class_loaded = True + if not class_loaded: + from mmcls.datasets.raf import FER_CLASSES + model.CLASSES = FER_CLASSES model.cfg = config # save the config in the model for convenience model.to(device) model.eval() diff --git a/mmcls/models/vit/vit_siam_merge.py b/mmcls/models/vit/vit_siam_merge.py index 77c8ebe..cc929cf 100755 --- a/mmcls/models/vit/vit_siam_merge.py +++ b/mmcls/models/vit/vit_siam_merge.py @@ -415,8 +415,8 @@ def init_weights(self, pretrained, patch_num=0): if patch_num != pos_embed.shape[1] - 1: logger.warning(f'interpolate pos_embed from {patch_pos_embed.shape[1]} to {patch_num}') pos_embed_new = resize_pos_embed_v2(patch_pos_embed, patch_num, 0) - else: # 去掉 cls_token - print('does not need to resize!') + else: # remove cls_token + print('does not need to resize!') pos_embed_new = patch_pos_embed del state_dict['pos_embed'] state_dict['patch_pos_embed'] = pos_embed_new diff --git a/resources/demo.jpg b/resources/demo.jpg new file mode 100644 index 0000000000000000000000000000000000000000..ea14281d04e36b525b6cfde42ae31b97660c350b GIT binary patch literal 6960 zcmbVwcT`i|vvwf#j!JI|sIQYH z*&N_5fR^U3{MEF7g%0>v)6>xb=@{r482%lM%uEc7%!~{SOsq`IEPsW1$HvOS_V?lM zBLA66OAiFnvoJC+{%i8TQfD0iZYIDhz$+k)IDnR$2FOiw)&mfu;-vo<9u@3=goYMK zN6)}Wg~3WSsN}>8XgR*1^>G06K2^bC+bb8O}epV-)wgAQzJSnMvYKRVR<} zFPxg7 z$H3^=_{8MY^vvus8nd#xw!X2sh2PshI3yeqk5B&cq5%N^LyM~aTVC8$UbKIG!1$LJ z4K0*9fZTNSmt+~vX+LJP^Exjs7s7PmPV(ofPG$*tV;s+O?_Vsuk_yY0@qelP%j|zf zEcE}x?Eew_|GZ`aY(N^S^MKp{5a9TT!l8Bs2#v9!j3R3PEY|FX9g4%0muA@H%1p?W zK8YDon@xd1T=(H_*=Dx9>6wqe@t!N=VDv{7A0CxLxHKMD1kCk)%#DNH513NRsrfn_ zjCeKltaW5{er07(Uj<7vbr!z(T{KIR7*NXlyoZs`q2%MTD;TDg1@E9 zjkHObW&%$n+3f9o`Ij)FO|-28M`&=A2DDUR7M1Bcx+GP&_G~JLw&E3nCLy#fk6?cq z?so==n{9<(!DF*hbzco8fs03e%THCP3LjOD=FUt}n5}U|w@0yu=45GC;gGvDlIod7 z#e|B$nOJYHgx+6+waNz$<^vFH)J-Nt-;R`vWW5}pa{E$DqMAs+tIZUJK8H& z_T0%RNzyvVWM3>2-GtMPLQmCh%h|`gs-CuzjlTjo9rs+eBg;(a@n7X3)!-`$dOz#3 zMVIED#g#dvy%7-hp?V+QM0)X=|c1#A4^5A|clFz1Z(xEcH<*#gRa*ravpD|y* zm%=UR!aHD28#V^h(MExccZs=Vl(Bn_7lckj;38ou&coc`Hu_S)e&^UZ#ta3 z#4O%gp_RPl43;nWEF$-)fi;uwGV$xn?Aw#{FZ>O^yS|#K#0P7>9uAfzNl-@0F{d7L zmTeq%56<<%H|T(g@Vp|I2LRVbF^G0@1#b*Hf#J6fPkq0Da`kH-QfaS}%Y7q)1u`Wt4w$ zR(;{h&IIjXM+ZMPW;D#CUOjF4e1inzr=v4~=5}l~pOs*#^g$?6U9jDKd$IRmU0z{*M1WWiCg`KN_@lmBeswN`C8c@9o{ z+$EpuqKs7fn9l&)W;A>8Oyx+ zs8h34{vejC?2>oXBky3+NetN`77ORWKWMhE_&MzA{p5V#JQt%DCg$*#;_FS7wh=Z5 z=_lYkt<+zB;b+aB9 zS}LB7-Yd>}ArzEK;V+r8A0pjpyso`MO@E+u#VXwB`aMtKY3F9_5Y7^;mY=HazSexs ztMSU}d~mh=wxy0-dhd_?#C)WEVt8neJc5lTt*8{fo&Z@|z zqTfxb(#hqq+v!bQWDHsMMj15Fqsmp9zz=JDxid3#=1sk(>%HbwdF*(kU!x) zyEy0gx*l0=*_$H{G9lLuymSFVRH@b$7rRYgADoK z?0Zxy5O-`JaLL)%rQ8Q5dLVO5uk{Ju|2XFHbUi+CI*M#rq{L>FzY^+2x&M$c8q3|) z`iU#hsd72Dauvc#=A`;Poq`foh?ztg6K9r5hD|2So?o~0T}HXQ!p#ZFo5w4H+mx$k zfX>@+01>2Z?J|gghC5kPp4fn++&}R4d3EEr4^AdUM@lFA$m1zBPP6_tm&Yn}f5>r} z?CjE|Xu6b>(*;A-1~x8uz3-Dp5zhdJ+j~W5J~EF`U;D+h+RD)OD=iZ@qmG~dR50Jk zx1;>{hr3UHAVp;bQ$ke7Fk^J5pl711O>`(iTwhK}Vx~ zU$t=OwAW{gb8frq#BVXQVh6hlewWlCyhCTUREafP+EhS9LzTJuSX%$EyzzdG@~qs( zLdwK={lK$OFs`f=je2V%9l)k*Uq9QI8YsCLKJ6)Cw|I4pz!-CUbekCFSFrUQ#26qt z)!lw+-0s|hD^wIh2tD+y$yi0){C(MNeQ+)%%tsmEh4Eo>keNT*qwVqa@x)C?OD@h88WmT+cn+pAWnh|L-}po_ zF6qO<4##rH^t$L`VlM8HCv!k-w^*$<{%KW-v-0TnN7HQ5mfA(g6|_V2IkU?WdP6Ng zcROyho|}2De|q~25M}|podUDK`@d)J`r7`}ea&>1^=6;lQQ`w$0TJSXz`?f(Ow4DN zDw6IG{1tr@@nLCO$KCJPAjKfRDHB{VX0`&GbQi(7IwC!av`4)uB7_kO>~LFyI!*24 zim_j@z9ea9OpLw*l&Pfh5>wN4Eys%-D=hobg$XB+n$CeU0CPVCvT1Wn;@(J_BO8z9 zC1>P~yGsJ{GtC98y=e}Qju@I~{?_{Wwi4zFdiM7-vG{!85FL+8tksec$=uYu!{HD*|MJdjHHEu>4W#rdO8 zb+pCEO0Q(;!rgH7pqNv}rWztQ=b!ze4J|I*ZkH#s)oph_Rx!4XkF#4J7kX^p-cTSJ zgKr6VNhp6*0W{7_Wv|7-bn|Xz=a(fJ3=UP<`u87`OMmn=Hg6UNnL}%fBE1X^Tk7H0 zV9M09<4hS38gOWivy++7_!FF)D%Kc zi`G0qtiLM2tRt{5uQVWV)7)LMEyeJC)hoFRR&psDUU90=Ier@%EmnF`Sjkoqjt`|D z*~TB~Kc(|Jerk`bfyA8CSKN&fy*za5T6(}D^SP2xeSfVtPMRv^@S4LIjg+EFADNQt zik_}4Q`U*!A#>*m-gx^nz=an?9^714;TFc+kllOq$Ez>D*K}JVOu;+SWiIOtJJBn3 zrekLSN2CZ@?DR`5Co9Yrm*o#VX^60^8hwOKv)SPZ7Z}G1A^Zz!rovK(gS*cFbkjs< z5Ee;71k?};A_O2EvyPvdA>mFLMst}jz4(3EXlMp%8zmAo;?~+DsCz}BbO(;)k+htV zwTYeVYWMJr-PQ{1?L@I*jY7Le$#+mkDp+bj4!qFmvWHo##VOJ-K8vlln69`HCvC&!-o;u25ZP3H9k zjm47S;2`OZGXQX_+D5XpTOeBlCh^+Dd6s}o1+lhk!5-n(-@wnioB=reWO!$OgQ99N zw8S83rQ_qhifwyFRFT$KQUCAF$uQ zz_?#IusgSlrc~h@2xIvEo?%G;$<>(4?Rd5?V&?*ax(q+hZoRq(`?{04eAn#>Q(cO)x}%9r^aq zQ>6l?(Un9gL;sc^hltO_@@3_>+mk;>US0m6o5H}8xPU^_2P1HZ?#BVpzIH)8M|MGZ zN&T1hm+}62>N2R191ei!A%6U$>wUTiMYm#pVC7Vzx9beT0Xgg1qxAdwsFo;^@W*xNe(dqMq*h_LN?mch8R_eOHx&5*4KW*pUdO(mLZPoly!}R$c|-)^z7QUM+-z%)C*McwagtlUy{&z! zw8P41y(bvo}@gvmq-6g)aa8{W7XsghYJBPHotFN~3ql$$R01#OaY>0s{ zZo*}8?CX)_yDuAfrB7aj@QA9Ho$rMEww~m6oB?uI z|2)mD)9t>YtDC%|`29`vSBP)3oeJdaewd}Mu6+&jkBPq;Y z_{^SJr%o9Gt$~|UrysB(?@C5{eq8ElyQ-(YXefknLZGM1405|yAaSbqNoCK~K7d4#DVA}B0Sk`1+F z-bXru-psD|f%(YL3ApgKmik250fL3d*{QZBj>CRpzn-G+!_R&hQzBQJob<^)!fv$; z@^zMHE!DL!s$4wg7N4YeNMaX~1G%&z6JNb`1`ugw$LT(7u@g%a2l)$N-t7O6Oy#6x) zYizH@=$&2+F7C0qx-I)yDK|0g>V}akZo6wTWqy7|H+u-sg$e857e~U7OB3u@^vFJV zmZe@&W3BY)m~8z89`$Z)8%M*e-jOh5wq9$i@E&@xR+1fR?^~} z!DI9X!}HfVjfFJ)zhCsS4nC~0?CMNP&BcI5za1Lh_2ZjKyPc$@eWMeTbp}w|P0NXt za#znIm=Q!y^RTT!xJ;e4rY6y#869s9X!a*4#q=O28C{zrD{chXx*+Phr%qu`#iv`# zu~q_H%$}w9L6m-3Eq$0y8qmDc0Mgm)g-4OK@L!uv7rZ*fes4p>;=6;?%Ef}xG1(6^ zEH!f>c=w*iy%!#a`+uKDjJQ_dz#&+e-6zeF&l;>kpV`+MmwcuqwvWv+Tmt!t@t?F} z-Am`q2Nh%lt!IDK>p^cSU(#e6(Z)iQiJr^5y)rX9`;T#)eT!lN+r5(yOD&O+ho^D} z!`GEk2BO6mbcQ&@jYmCi_eN+x4#_ci6lO%q9v{amzB5~@pkOu;yp)k<-sED%#W0w8 zVXv>{hrIT9!`uf>Ifqj)KB5!iwYM)S+Tt{hTbed$81}&@r^P1*zfZazaJTQorU9K8 ziL7HU$5v!>z07-!@lKH7z<=+2@9M1<5*HUX+JXd7I@7DRPF@hV2@$fA&3?VZJx@o! zXo+(dOIam75x7v*4-}E_?RtA!)Je<Gw> z!1M_2cO)uRset1=jOsB%30|nyY9=T7PXr3zheZCCm>@Hw zEkVKojwZr{6!@O_G)GZr`34g|>X7FIOj05{jJ5_X$AY+=kKROMOeF2&zfZ4--Qlz# z4_EuRdmBF-n7<~|Kcwk%Q2F9CO-pWTkqzc|IgFcN*d>DPr?D5Any2Qn)>gRKY= zCo!V-&-P+na9TNhD2IP47t2y2@NR`pK-d(-0Us2UiL9eRcwTc>X3t-uUmp`Q$SpBl zYj&C#Oeivrlw1~?3vvPDJeOMoMc;T-u8{7}4`w(@F)*lf@B@NO*`Wq?KX(grT}nm0 zC_I~~N}0>SB0VEtCBRm`yO30BC|gt#r5!VHV4d-H0GScbVu;Ch1hOBlmSw21msZNz z2hT;C>~WNE>?fu|ovPCOKc&s!4y(q}^e4X_X&F+Fzz%$ocd?eQPX*+}8PWU&wG06+ z>iHQq+7AUrl+ncFQP)`B$BD{euRh{!zlvLJ?q+HF0c4v$5u>)spIdY)eGWd?BC02u zqOX-tQJl7%o9nf0V0v>66Ef8xnhoaJ?R)pqBu*Umr(~c_S5@SF2ZF;YAhN$18ky(7 z3JtyhWsM8hO@2~^iDR2Oq&A@F7I2JYX_*;NR}Leuql80&_S%(>PkK0=J>M->dEq0} zBr+XEyb6iL9_}X2sx(I5M2yO^-F#aX|E(l!OPS{+@St)FM6f&k;_Gg>jE$UY)gavK dQ7W&^oC3I;gwl{cyMBC17%G_`iP1fq`7iLm(op~a literal 0 HcmV?d00001